AI Science Tracker

AI discoveries in chemistry

111 discoveries from 2020 to 2026

Discoveries

Foundation model and generative AI yield 3CL protease inhibitors confirmed at sub-micromolar IC50

An AI-assisted 3CL-protease inhibitor study selected 249 molecules from about five million generated candidates and evaluated 18 using free-energy perturbation. Five molecules were experimentally tested; all five inhibited the target below 1 micromolar IC50, and one inhibited more strongly than its lead. MMELON activity predictions had Pearson R=0.72 in a blind test. This summary is limited to the abstract and November 13, 2026 publication metadata; it does not establish antiviral efficacy in people.

Technology: MMELON biomedical foundation model with generative molecular models

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Five synthesized molecules tested for 3CL-protease inhibition after computational candidate generation and filtering.
  • Still unknown: Source checking does not establish independent confirmation. Review date: 2026-11-15.
  • AI role: Generative models produced candidate molecules; the MMELON foundation model predicted activity and supported downselection.

Evidence sources: https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=EXT_ID%3A43024107&resultType=core&format=json

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review scope and date are not recorded; source checked does not establish full-text review.

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Machine learning guides discovery of selective submicromolar CK2 inhibitors

The accepted-paper abstract reports that docking, CDK-liability filtering, machine learning and error prediction guided more selective CK2 inhibitor leads. The authors report submicromolar hits and a 90% pipeline hit rate; kinase-assay figures support biochemical inhibition. These are discovery-stage compounds, with no demonstrated clinical benefit. The hit rate describes the authors' selected pipeline, not a general drug-discovery success rate.

Technology: Machine learning; prediction-error modeling; molecular docking

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Kinase-assay figures report CK2 inhibition by selected hit compounds.
  • Still unknown: Full study methods, independent reproduction and clinical outcomes were not assessed.
  • AI role: ML and error prediction prioritized compounds after docking and off-target filtering.

Evidence sources: https://www.nature.com/articles/s42004-026-02268-0 · https://www.nature.com/articles/s42004-026-02268-0_reference.pdf

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Editorial assessment: Impact 2/5 · Validation 3/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

DeltaCata guides methyltransferase engineering with improved catalytic efficiency in vitro

The accepted-paper abstract reports that DeltaCata, a geometric graph neural network predicting mutation effects on enzyme kinetics, guided engineering of methyltransferase sGmOMT2. The authors report up to a sevenfold improvement in catalytic efficiency for in vitro icaritin synthesis. This experimental application goes beyond the model benchmarks; industrial-scale performance is not demonstrated by the reviewed abstract.

Technology: DeltaCata; Siamese geometric graph neural network

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: The abstract reports experimental enzyme engineering and an in vitro catalytic-efficiency improvement.
  • Still unknown: Full experimental methods, industrial performance and independent replication were not assessed.
  • AI role: The model predicts mutation-induced kinetic changes to prioritize enzyme variants.

Evidence sources: https://www.nature.com/articles/s41467-026-79287-x

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Editorial assessment: Impact 2/5 · Validation 3/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Google DeepMind

Agentic cheminformatics: new macrocycles from literature analysis

An agent using Cheminformatics Skills for Antigravity collected thousands of compounds analogous to crizotinib and lorlatinib, analyzed 568, and proposed new rigidified macrocycles with synthesis routes in about an hour. The proposals are computational hypotheses that still need to be synthesized and validated.

Technology: Google Antigravity, Cheminformatics Skills

Publication: Announcement only · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Announcement only
  • Test type: retrospective or benchmark
  • Tested: The source describes computational evaluation only: benchmarks of several machine learning architectures on four readouts with scaffold splits, and predicted potency; no synthesis or experimental validation is reported.
  • Still unknown: Every proposed design must still be synthesized and validated; the data are sparse and assay noise is high; the designs are testable hypotheses, not finished drugs.
  • AI role: An agent searched literature for crizotinib and lorlatinib analogues, curated 568 compounds, ran matched molecular pair analysis, benchmarked models, and proposed macrocycles and synthesis routes.

Evidence sources: https://deepmind.google/blog/agentic-cheminformatics-with-google-antigravity/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: blog. Paper: not linked in this catalog; this does not mean no paper exists.

Review scope and date are not recorded; source checked does not establish full-text review.

Editorial assessment: Impact 2/5 · Validation 2/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

KU Leuven

AI agent maps proton-gated water loss in perrhenate-catalysed deoxydehydration

The paper reports a computational mechanism for perrhenate-catalysed deoxydehydration: proton-assisted water loss enables ring closure and alcohol reduction. Under human-defined hypotheses, a Claude Code agent located transition states, which the authors validated. Coupled-cluster-anchored microkinetic modeling predicted a rate increase of about eight orders of magnitude over the best proton-free pathway. At matched conversion conditions, modeled barriers were consistent with the 31.5–34.0 kcal/mol window inferred from previously reported experimental turnover; this was not a new wet-lab demonstration.

Technology: Claude Code LLM agent for human-directed reaction-network and transition-state search, combined with electronic-structure calculations and microkinetic modeling

Family Claude: profile · models page

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Agent-located transition states verified by the authors; computed energetics, reaction-network and microkinetic consistency checks; comparisons to existing turnover-derived barrier estimates.
  • Still unknown: Main paper reviewed; supplementary files were not read. The mechanism and rate changes are predictions compared with prior experimental kinetics, not a new experimental confirmation.
  • AI role: Claude Code carried out calculation/search/triage workflows for author-specified mechanistic hypotheses; authors validated the located stationary points.

Evidence sources: https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=DOI%3A%2210.1039%2Fd6sc06865k%22&resultType=core&format=json · https://www.ebi.ac.uk/europepmc/webservices/rest/PMC13720772/fullTextXML

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review scope and date are not recorded; source checked does not establish full-text review.

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Machine-learning-guided photoredox design experimentally enables three transformations

A workflow combining a variational-autoencoder latent space of 11,069 literature photoredox reactions with random-forest catalyst/reagent recommendations was used to propose three transformations. Experiments synthesized a previously inaccessible substituted dihydroquinolinone 43 using fac-Ir(ppy)3, a related oxidative cross-coupling product 46 in 64% yield, and heterocycle product 49 in 74% yield. The paper presents these as experimentally validated transformations identified through model-assisted precedent retrieval and condition selection.

Technology: Variational autoencoder for reaction-space navigation and random-forest multilabel catalyst/reagent recommendation

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Experiments synthesized a previously inaccessible substituted dihydroquinolinone 43 using fac-Ir(ppy)3, a related oxidative cross-coupling product 46 in 64% yield, and heterocycle product 49 in 74% yield.
  • Still unknown: The work is a human-guided photoredox design workflow, not autonomous synthesis. The three examples were selected by researchers and use a literature-trained model; the paper establishes feasibility for these tested transformations, not general transfer or broad yield improvement. Reported 64% and 74% yields apply only to the specific demonstrated products; further optimization and substrate-scope work are needed.
  • AI role: The VAE organized 11,069 literature-derived reactions to retrieve traceable precedents; random-forest models recommended catalyst and reagent classes for proposed reactant-product pairs. Chemists used the model-suggested conditions to test three transformations, including two modifications beyond the reported precedent and one previously undocumented photoredox synthesis, then measured isolated product yields.

Evidence sources: https://pubmed.ncbi.nlm.nih.gov/42953598/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

LLM-based virtual screening identifies two NIR-II fluorophores tested in vitro and in vivo

Researchers trained a large-language-model framework to predict fluorophore emission wavelengths and used virtual screening to prioritize NIR-II imaging molecules. Two candidates were experimentally characterized and showed favorable imaging performance in vitro and in vivo, according to the abstract. The abstract supplies no quantitative imaging effect sizes. This summary is limited to the abstract and October 15, 2026 publication metadata; the candidates are imaging probes.

Technology: Large language model for structure-property prediction, with virtual screening

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Two computationally prioritized fluorescent candidates characterized for imaging in vitro and in vivo.
  • Still unknown: Source checking does not establish independent confirmation. Review date: 2026-11-15.
  • AI role: A large language model predicted emission wavelengths and guided virtual screening to choose candidates.

Evidence sources: https://onlinelibrary.wiley.com/doi/10.1002/smll.76251?af=R

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review scope and date are not recorded; source checked does not establish full-text review.

Editorial assessment: Impact 2/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Lila Sciences

Catalysts without iridium or ruthenium for acidic oxygen evolution

An AI-guided, human-supervised platform assessed 2,942 catalysts and identified complex palladium oxides. InMnPdOx maintained an overpotential below 0.5 V for more than 1,000 hours in sulfuric acid at 10 mA/cm². This is a preprint and laboratory result, without validation in a PEM electrolyzer.

Technology: Sequential learning, composition–property models, and LLMs

Publication: Preprint · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Preprint
  • Test type: direct test
  • Tested: Lead catalysts were tested in 1 M H2SO4 at 10 mA/cm2: InMnPdOx kept overpotential below 0.5 V for over 1,000 hours; a retrospective benchmark compared the agent with Bayesian optimization and in-context language-model selection.
  • Still unknown: Testing in a full PEM electrolyzer is not reported, and 'stable' means overpotential under 0.5 V rather than a broader durability assessment.
  • AI role: A sequential-learning agent with ML composition models, adaptive optimization and LLM reasoning ran a human-supervised closed loop with over 90% automation.

Evidence sources: https://arxiv.org/abs/2609.30133

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review scope and date are not recorded; source checked does not establish full-text review.

Editorial assessment: Impact 4/5 · Validation 3/5 · Novelty 4/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

RetroChimera: predicting small-molecule synthesis

Published in Nature, the method combines a Transformer model (R-SMILES 2) and a template-based graph model (NeuralLoc) through a learned vote. In blind tests with expert chemists, 90% of complete routes were accepted, compared with 20–50% for the component models or a baseline, and nine of ten difficult targets were solved. Evaluation covered only ten targets; one wrong prediction invalidates an entire route.

Technology: RetroChimera (Transformer and template-based model)

Family RetroChimera: profile · models page

Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: retrospective or benchmark
  • Tested: Blind tests by PhD-level chemists rated individual steps and complete routes for ten difficult targets: 90% of RetroChimera routes were accepted, and nine of ten targets were solved, versus 20-50% for the component models or baseline.
  • Still unknown: The route-level evaluation covers only ten targets; one incorrect prediction invalidates a route; NeuralLoc is constrained on reactions outside its template library, and R-SMILES 2 may hallucinate.
  • AI role: An ensemble of a Transformer (R-SMILES 2) predicting precursors and a graph neural network (NeuralLoc) selecting templates, combined by a learned rank-dependent vote.

Evidence sources: https://www.microsoft.com/en-us/research/blog/improving-synthesis-prediction-of-small-molecules-at-scale-with-retrochimera/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: blog. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Editorial assessment: Impact 3/5 · Validation 2/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Bayesian optimization finds mild solvent-free mechanochemical conditions for cyclopropanation and epoxidation

A Bayesian-optimization-guided mechanochemical study identified conditions for the established Johnson–Corey–Chaykovsky reaction using KOH in place of hazardous NaH, under air-equilibrated conditions and with no solvent or liquid-assisted grinding agent. In the chalcone benchmark, one optimized condition reached cyclopropanation completion in five minutes; the study also demonstrated cyclopropanation and epoxidation across multiple substrates, including yields of 45–75% for four additional substrates.

Technology: Bayesian optimization implemented with BoTorch/PyTorch, including multi-objective optimization of product yield and process mass intensity

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: A Bayesian-optimization-guided mechanochemical study identified conditions for the established Johnson–Corey–Chaykovsky reaction using KOH in place of hazardous NaH, under air-equilibrated conditions and with no solvent or liquid-assisted grinding agent.
  • Still unknown: The Johnson–Corey–Chaykovsky transformation itself is established chemistry; the finding is a measured, safer and faster mechanochemical process window rather than discovery of a new reaction class. A 5-minute completion time and other yields apply to the tested substrates/conditions; some prior solvent-minimized conditions could not be reproduced. The scale and substrate scope remain those reported in the paper.
  • AI role: The BO models selected among six reaction variables (base identity/loading, time, milling frequency, liquid-assisted grinding agent and loading) using experimental yields and process-mass-intensity outcomes. Researchers ran the experiments, performed synthetic substrate-scope tests and used time-resolved in-situ X-ray diffraction to study milling effects.

Evidence sources: https://www.nature.com/articles/s41467-026-75499-3

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

An AI system improves a Chan–Lam coupling reaction

The system identified TEMPO as a useful additive for coupling primary sulfonamides; chemists reproduced yield improvements in 11 of 14 substrate pairs at bench scale. The work retains human supervision and specialized infrastructure; independent reproducibility and generalization remain to be checked.

Technology: GPT-5.4 and Maria

Family GPT-5: profile · models page

Publication: Announcement only · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Announcement only
  • Test type: direct test
  • Tested: Over two cycles in an automated lab, yields improved for 88% of boronic acids and 83% of sulfonamides tested (mean 16.6% to 25.2%); human chemists repeated representative reactions at bench scale, with higher yields for 11 of 14 substrate pairs.
  • Still unknown: Not stated in the reviewed material
  • AI role: GPT-5.4 connected to Molecule.one Maria proposed, designed and analyzed experiments; humans steered prompts, selected proposals, made limited corrections and validated the final result.

Evidence sources: https://openai.com/index/ai-chemist-improves-reaction/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: press release. Paper: not linked in this catalog; this does not mean no paper exists.

Review scope and date are not recorded; source checked does not establish full-text review.

Editorial assessment: Impact 4/5 · Validation 3/5 · Novelty 4/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Deep learning finds multimetallic single-atom catalysts across material families

The Crossbreeding Neural Network (CBNN), trained jointly on carbon-supported single-atom catalysts and perovskite oxides, screened 8,008 candidates. Its top W-Mo-Ru-Rh design on a perovskite support outperformed the comparison catalysts in electrochemical tests, and the activity ranking of 12 tested catalysts matched the predictions. These counts come from the IBS institutional release; the paper abstract confirms the experimental result but does not quantify the activity gain. The catalog date is the Nature Materials publication date, 28 May 2026.

Technology: CBNN (Crossbreeding Neural Network)

Publication: Announcement only · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Announcement only
  • Test type: direct test
  • Tested: The AI-predicted catalysts were synthesized and tested for alkaline oxygen evolution; the activity ranking of 12 tested catalysts matched the predictions.
  • Still unknown: The comparison is limited to catalysts synthesized in this study; use in batteries, energy storage and drug discovery is not demonstrated.
  • AI role: A Crossbreeding Neural Network trained on carbon-supported single-atom catalysts and perovskite oxides screened 8,008 multimetallic candidates.

Evidence sources: https://www.ibs.re.kr/cop/bbs/BBSMSTR_000000000738/selectBoardArticle.do?nttId=26725&pageIndex=1&searchCnd=&searchWrd=

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: press release. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Editorial assessment: Impact 4/5 · Validation 3/5 · Novelty 4/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

AI generates compounds with experimentally validated pancreatic cancer cell selectivity

The IRB Barcelona team trained cell-line cytotoxicity predictors on data from more than 11,000 compounds assayed across pancreatic-cancer and control cell lines, then integrated the models with a reinforcement-learning generative system to design molecules for cell-selective effects. In follow-up tests, 45 AI-designed molecules or close analogues were assayed; 20 showed the intended cell-specific cytotoxicity, and dose-response testing confirmed 18 of those 20. Across the seven selectivity exercises, at least one hit was found in five; measured IC50 values could be calculated for 12 confirmed hits and ranged from 0.97 to 16.6 μM in the targeted cell lines.

Technology: Cell-line-specific cytotoxicity prediction models coupled to a reinforcement-learning molecular generator (REINVENT-based workflow)

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: The IRB Barcelona team trained cell-line cytotoxicity predictors on data from more than 11,000 compounds assayed across pancreatic-cancer and control cell lines, then integrated the models with a reinforcement-learning generative system to design molecules for cell-selective effects.
  • Still unknown: This is a proof-of-concept hit-discovery study using in-vitro cell-line assays, not evidence of tumor selectivity in patients or an animal efficacy result. Only binary cytotoxicity labels at 10 μM informed the predictors; 45 generated structures or close analogues were tested, with 20 showing desired phenotypes and 18 confirming in dose-response assays. Exact generated structures were not always the synthesized compounds, and measured IC50 spans 0.97–16.6 μM for 12 of the confirmed hits. The paper explicitly characterizes this as an early hit-discovery stage.
  • AI role: The authors trained predictive models on phenotypic screening data from >11,000 compounds, then used their scores as rewards/objectives in a generative workflow to propose molecules intended to kill one pancreatic-cancer cell line while sparing another or control cells. Candidate molecules or close purchasable analogues were experimentally tested and dose-response validated.

Evidence sources: https://pubmed.ncbi.nlm.nih.gov/42162204/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Joint molecular modeling identifies seven low-micromolar kinase hits beyond its training chemistry

A joint molecular model combined bioactivity prediction and molecule reconstruction to estimate how unfamiliar a candidate structure was to its training distribution. In a prospective screen of commercial compounds against PIM1 and CDK1, unfamiliarity-informed selection led to seven compounds with low-micromolar potency and limited similarity to training molecules. The authors also evaluated the metric across 33 experimentally labelled bioactivity datasets.

Technology: Joint molecular model (JMM): semi-supervised molecular property-prediction and reconstruction model using a SMILES encoder; reconstruction loss defines the unfamiliarity score

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: The authors also evaluated the metric across 33 experimentally labelled bioactivity datasets. The reported wet-lab candidate assays directly test the molecular-model predictions within the study.
  • Still unknown: The kinase finding is a prospective in-vitro hit-identification result, not evidence of therapeutic efficacy. Seven low-micromolar hits were reported across PIM1 and CDK1; CDK1 was described as independent of the study’s earlier dataset analysis. The study screened only a small selected set per target after computational ranking. It does not establish the unfamiliarity measure as universally superior or demonstrate animal or clinical benefit.
  • AI role: The model used reconstruction-based unfamiliarity alongside predicted bioactivity and prediction uncertainty to rank molecules for a prospective virtual screen. Researchers tested selected commercial compounds against PIM1 and CDK1; seven compounds showed low-micromolar potency. AI supported prioritization and selection, while wet-lab assays established activity.

Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC13109061/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Editorial assessment: Impact 2/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

MetGenX annotates two previously uncharacterized metabolites in mouse liver

MetGenX, a structure-informed neural-network workflow for generating structures from tandem mass spectra, was applied to untargeted mouse-liver metabolomics and reported two previously uncharacterized metabolites, adenosyl-cysteine and adenosine 5′-diphosphate 5′-inosine (Ap2I). The paper reports validation against chemical standards using retention time and MS2 spectral matches.

Technology: MetGenX structure-informed encoder-decoder neural network with LightGBM reranking

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: MetGenX, a structure-informed neural-network workflow for generating structures from tandem mass spectra, was applied to untargeted mouse-liver metabolomics and reported two previously uncharacterized metabolites, adenosyl-cysteine and adenosine 5′-diphosphate 5′-inosine (Ap2I).
  • Still unknown: The original Nature full-text route was inaccessible in this session; the primary-source PubMed article page provides the paper abstract and figure captions describing standard-based retention-time/MS2 matches for the two structures. The finding is metabolite annotation in mouse tissue, not evidence of human biological function. Independent replication was not checked.
  • AI role: The model generated and reranked candidate molecular structures from MS2 spectra; researchers applied a multi-step workflow to mouse-liver samples and compared the two reported metabolite candidates with standards. AI proposed annotations, while standard-based chromatographic and spectral matching supported the assignments.

Evidence sources: https://pubmed.ncbi.nlm.nih.gov/42009661/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Editorial assessment: Impact 2/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Catalysis AI Agent derives design rules for copper single-atom alloys for CO2 electroreduction

An LLM-based Catalysis AI Agent identified an energy descriptor for classifying copper-based single-atom alloys and a structural descriptor for predicting activation energy. The team reports experimental testing and that the approach extends to other metal dopants. Evidence caveat: the press release gives no quantitative activity or selectivity values and does not name the experimental group, so validation is limited.

Technology: Catalysis AI Agent (LLM-based agent)

Publication: Announcement only · Test context: Test context unspecified in annotation. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Announcement only
  • Test type: not reported
  • Tested: The press release says the approach was tested experimentally and extends to other metal dopants, but the experiments are not described.
  • Still unknown: Not reported; the release gives no quantitative activity or selectivity values and names no experimental group.
  • AI role: An LLM-based Catalysis AI Agent trained on the DigCat database mined trends and identified an energy descriptor and a structural descriptor.

Evidence sources: https://www.wpi-aimr.tohoku.ac.jp/en/achievements/press/2026/20260326_002166.html

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: press release. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Editorial assessment: Impact 3/5 · Validation 1/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Machine learning identifies covalent organic frameworks for hydrogen peroxide photosynthesis

An ensemble machine-learning framework combined data augmentation, multilevel structural descriptors and physics-informed features to screen more than 10,000 covalent organic framework candidates. The researchers synthesized and tested eight selected COFs. The paper reports COF-343 at 12,978.7 μmol H2O2 h−1 g−1, with additional measured results for COF-2449, COF-2835 and COF-8484. This is one study-level materials result, not four discoveries. Detailed standard-test conditions in the supplement were not reviewed, and the results do not establish industrial performance. A data/code deposit may predate the journal article, but its date and result content were not checked here.

Technology: Deep neural network ensemble; random-forest band-position model

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: The indexed primary article reports synthesis and photocatalytic testing of eight selected COFs, including measured H2O2 production rates for COF-343, COF-2449, COF-2835 and COF-8484.
  • Still unknown: Detailed standard-test conditions in the supplementary information were not reviewed; the reported laboratory rates do not establish industrial performance. Zenodo deposit chronology/content was not checked.
  • AI role: A DNN ensemble predicted photocatalytic yields and a random-forest model estimated band positions to guide COF screening.

Evidence sources: https://www.nature.com/articles/s44160-026-01037-0

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review date: 16 Nov 2026. Review scope: Primary Nature Synthesis article page, abstract, methods/results text and author metadata as indexed by publisher..

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Adaptive AI experiment planning finds multi-promoter catalysts in under 50 experiments

Fritz Haber Institute, with BASF and BasCat, used adaptive experiment planning in a self-driving lab to search over 10^13 promoter combinations with fewer than 50 experiments. The method was validated on propane-to-propylene conversion, and the authors report synergistic promoter effects that explain the catalyst's performance. The claim of outperforming the industry reference is not quantified in the release, so check the paper for numbers.

Technology: Adaptive experiment planning (Bayesian-style, gray-box) in a self-driving lab

Publication: Announcement only · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Announcement only
  • Test type: direct test
  • Tested: The method was validated on propane-to-propylene conversion, and the release says it identified a catalyst superior to the industry reference, without figures.
  • Still unknown: No quantitative results are given and the release states no explicit limitations.
  • AI role: Adaptive AI planned experiments in a self-driving laboratory, using a gray-box strategy that also explained the effects of individual promoters.

Evidence sources: https://www.fhi.mpg.de/2217409/2026-03-24_ACS_Catal_SDLs

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: press release. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Editorial assessment: Impact 4/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

ChatHEA assistant guides high-entropy alloy catalysts for fuel cells

ChatHEA, a domain-specific LLM assistant for high-entropy alloy electrocatalysis, guided synthesis and screening of 100 five-element alloy catalysts. The FeCoCuPtIr catalyst reached a peak power density of 0.789 W cm-2 in a fuel cell and outperformed commercial Pt/C in electrochemical tests. These figures are reported by the institutional press release; its numeric activity values and DOE target are not given. The catalog date is the paper's 14 March 2026 publication date, stated in the 8 July press release; the full paper was not reviewed.

Technology: ChatHEA (LLM-based domain assistant)

Publication: Announcement only · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Announcement only
  • Test type: direct test
  • Tested: 100 five-element alloy catalysts were screened by high-throughput experimentation; FeCoCuPtIr outperformed commercial Pt/C in electrochemical tests and reached 0.789 W cm-2 peak power density in a fuel cell.
  • Still unknown: Not reported; the page gives no detailed durability data and lists no limitations.
  • AI role: ChatHEA, a domain-specific LLM assistant, supported literature review, element design, experiment planning, data processing and activity analysis.

Evidence sources: https://www.wpi-aimr.tohoku.ac.jp/en/achievements/press/2026/20260708_002238.html

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: press release. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Editorial assessment: Impact 4/5 · Validation 3/5 · Novelty 4/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Bayesian optimization identifies enzyme nanohybrids with high activity recovery

Parallelized hybrid-space Bayesian optimization prioritized metal/ligand carriers and conditions for enzyme immobilization. The primary abstract reports activity recovery of 100% for glucose oxidase, 90% for catalase and 79% for Candida antarctica lipase B across the demonstrated workflow. These are enzyme-specific assay outcomes, not discovery of new enzymes or proof of industrial stability.

Technology: Parallelized hybrid-space Bayesian optimization (PHBO)

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: The primary source reports the study-specific measured result summarized here.
  • Still unknown: Independent replication and clinical or industrial efficacy are not established.
  • AI role: The optimizer proposes carrier compositions and experimental conditions, then experimental activity feedback informs later proposals.

Evidence sources: https://www.nature.com/articles/s41467-026-70251-3

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review date: 16 Nov 2026. Review scope: Primary source indexed full text or abstract and publication metadata reviewed; scope stated in source_checks..

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Generative design and screening produced experimentally tested antimicrobial QACs

A topology-aware generative workflow and activity predictor prioritized quaternary ammonium compounds (QACs) for synthesis and testing. Across two workflows the authors report 29 selected compounds characterized and 11 QACs designated as novel, with measured minimum inhibitory concentrations of 1–32 μM against bacterial pathogens. Earlier same-team conference disclosures had already described AI-generated QAC design and synthesis; this record covers the later paper’s bounded tested results and does not claim the first AI-generated QACs or universal chemical novelty. Activity and cytotoxicity remain assay-specific and do not establish clinical use.

Technology: Topology-aware QAC generative model with structure/activity prediction and filtering (Q-HERO)

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: The article reports synthesis and antimicrobial testing across two workflows, with 29 selected compounds characterized and 11 reported novel QACs with MIC values of 1–32 μM.
  • Still unknown: Prior same-team AI-QAC design and synthesis were described in 2022 and 2024 institutional conference abstracts. The measured activity is limited to the reported laboratory assays and does not establish clinical efficacy.
  • AI role: A topology-aware generative model proposed QAC structures, and learned activity filtering helped prioritize candidates for synthesis and antimicrobial assays.

Evidence sources: https://pubs.acs.org/doi/10.1021/acs.jcim.6c00390 · https://pubs.acs.org/jcisd8/article/66/6/3117/5080435/Topology-Aware-Generation-and-Activity-Based

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review date: 17 Nov 2026. Review scope: Primary ACS indexed full article, publication history, methods/results, named affiliations and indexed supporting-information description; earlier team conference disclosures as identified in the frozen source review..

Editorial assessment: Impact 2/5 · Validation 3/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Machine learning guides a low-precious-metal RWGS catalyst

An iterative machine-learning-guided campaign optimized precious-metal and additive compositions and support over 35 prediction–experiment cycles. The authors report a low-loading Pt–Ru–Pd catalyst that outperformed a commercial benchmark and their earlier high-PGM design in the reported reverse water–gas shift assay. This is a laboratory catalyst result, not demonstrated industrial performance.

Technology: Machine-learning-guided iterative catalyst composition and support optimization

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: The publisher article reports the 35-cycle campaign and its best low-PGM catalyst outperforming the stated commercial and earlier high-PGM comparators.
  • Still unknown: The result is specific to the reported assay; scale-up and industrial operation are not established.
  • AI role: Machine-learning-guided predictions prioritize PGM/additive compositions and support for iterative experimental testing.

Evidence sources: https://pubs.acs.org/doi/10.1021/acscatal.5c09278

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review date: 16 Nov 2026. Review scope: Publisher indexed primary article text including author affiliations, article history and abstract/result; full-text access remains publisher-limited..

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

LFAST: an LLM agent drives robotic synthesis of the new COF-2000 framework

A GPT-4o-based agent (LFAST) screened reaction conditions over a 96-condition matrix and refined them after each round, raising the crystallinity index of a reference covalent organic framework by 350% and yielding COF-2000, a previously unreported material with higher structural order. The reported validation is crystallinity measurement of the synthesized products; the authors do not pursue applications. Publication date is taken from secondary reports of the JACS paper and should be checked against the journal page.

Technology: LFAST (LLM For Accelerated Synthesis Technique), GPT-4o agent with robotic experimentation

Family GPT-4: profile · models page

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Products of LFAST-guided conditions characterized by powder X-ray diffraction: crystallinity index raised by 350% on the benchmark COF TpPa-SO3H; COF-2000 characterized by PXRD and crystallinity index only.
  • Still unknown: Evidence rests on one benchmark system and one new COF; the full text was not read, so limitations discussed there are unknown.
  • AI role: LLM deep research agent mined the literature for synthesis conditions; an automated platform ran them and PXRD analyzed the products.

Evidence sources: https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=DOI:10.1021/jacs.5c23233&resultType=core&format=json · https://bidmap.berkeley.edu/news/new-ai-agent-paradigm-shift-cof-synthesis

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: press release. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 4/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Machine-learning and expert co-screening identifies two new silver-catalyzed cycloadditions

Researchers at the Zelinsky Institute used unsupervised machine learning to filter and cluster candidate atom-economic cycloaddition reactions generated from quantum-chemical data, then used a digital co-expert and chemist review to prioritize experiments. The team experimentally confirmed two previously undescribed silver-catalyzed cycloaddition reactions yielding new products within about one week.

Technology: Machine-learning candidate filtering and clustering combined with the digital co-expert reaction-prioritization workflow

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Researchers at the Zelinsky Institute used unsupervised machine learning to filter and cluster candidate atom-economic cycloaddition reactions generated from quantum-chemical data, then used a digital co-expert and chemist review to prioritize experiments.
  • Still unknown: Two reactions were confirmed from a heavily filtered set of computational candidates, with final selection and validation by chemists. The reported approximately 180-fold acceleration is a comparison for the expert-screening bottleneck, not a general chemical-discovery speedup. The abstract does not specify individual products or reaction yields; no such details are inferred here.
  • AI role: Unsupervised ML filtered and clustered computationally generated reaction candidates to reduce the set for review; the digital co-expert helped prioritize candidates. Human chemists made the final selection and performed experimental validation. The source reports two newly confirmed cycloadditions; the system was a human-AI pipeline, not autonomous discovery.

Evidence sources: https://onlinelibrary.wiley.com/doi/10.1002/ange.202523905

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

DeepMet anticipates and helps identify dozens of previously unrecognized mammalian metabolites

DeepMet, an LSTM chemical language model trained on known human metabolites, generated a ranked space of metabolite-like structures and was integrated with mass spectrometry data. Researchers report identifying several dozen previously unrecognized mammalian metabolites across standards, human biofluids, mouse tissues and repository data. Some standard-supported assignments still have isomer ambiguity; the study also found known metabolites missing or misannotated in HMDB.

Technology: DeepMet chemical language model (three-layer LSTM over SMILES), integrated with MS/MS-spectrum prediction and meta-learning

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Researchers report identifying several dozen previously unrecognized mammalian metabolites across standards, human biofluids, mouse tissues and repository data.
  • Still unknown: A subset of assigned structures retain regioisomer/stereoisomer ambiguity; human oversight helped prioritize structures for synthesis. The model explores chemical space similar to its training set and may miss divergent biosynthetic chemistry. The paper distinguishes novel metabolites from known compounds missing in HMDB.
  • AI role: Learned metabolite-like chemical structures and prioritized structures likely to exist; combined with accurate mass/MS/MS and reference standards to support identification of previously unrecognized metabolites.

Evidence sources: https://www.nature.com/articles/s41586-025-09969-x.pdf

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

DrugCLIP screens about 10,000 human proteins against 500 million compounds

The January 2026 Science abstract reports DrugCLIP screening roughly 10,000 human proteins against 500 million compounds. Experimental hits included 15% for the norepinephrine transporter, with structures of two inhibitor complexes, and 17.5% for TRIP12 using AlphaFold2 structures. Most database results remain computational. The work also has a bioRxiv preprint first indexed in September 2024.

Technology: DrugCLIP (deep contrastive learning)

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Biochemical hit testing and structures of two NET inhibitor complexes.
  • Still unknown: Not assessed: full assay denominators, uncertainty and genome-wide experimental coverage.
  • AI role: Contrastive learning ranked protein-compound pairs.

Evidence sources: https://pubmed.ncbi.nlm.nih.gov/41505557/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Editorial assessment: Impact 4/5 · Validation 3/5 · Novelty 4/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

PROBY screening uncovers target-selective fluorescent probes for tau, BCL-2 and TDP-43

The PROBY deep-learning pipeline screened 26,416 target-validated molecules for useful optical properties, identified 11 candidates for testing and found target-responsive probes PE859 (tau), obatoclax (BCL-2) and B3 (TDP-43). The study reports selectivity assays, cell/tissue labeling and ex-vivo tumor imaging with obatoclax. AI-guided structural modification of PE859 yielded 859-2, which enabled in-vivo two-photon and cleared-brain imaging of tau pathology in P301S transgenic mice; its measured tau-aggregate dissociation constant was 54.8 nM.

Technology: PROBY two-stage message-passing neural network pipeline (PROBY-L1 fluorescence classification and PROBY-L2 photophysical-property prediction)

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: The PROBY deep-learning pipeline screened 26,416 target-validated molecules for useful optical properties, identified 11 candidates for testing and found target-responsive probes PE859 (tau), obatoclax (BCL-2) and B3 (TDP-43).
  • Still unknown: The PROBY hits are repurposed molecules from a target-validated library, not newly synthesized compounds. Target/probe performance is assay-specific; tau in-vivo imaging was demonstrated in a small transgenic-mouse study (n=3), not humans. This entry describes probe function and imaging results, not treatment effects or diagnostic performance in patients.
  • AI role: PROBY ranked target-validated compounds by predicted fluorescence and seven optical properties, enabling selection and experimental verification of target-specific probes; the team then used AI-predicted structural modification alongside synthesis and bioassays to optimize PE859 into 859-1/859-2.

Evidence sources: https://pubmed.ncbi.nlm.nih.gov/41441857/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Diffusion-guided design tests bimetallic ammonia-decomposition catalysts

A diffusion model guided by a learned nitrogen-adsorption predictor generated bimetallic candidates near the target adsorption energy. Co75Mo25 was among the generated candidates and maps exactly to the tested Co3Mo1 ratio; under the reported supported-catalyst assay it reached 18.6 μmol gmetal⁻¹ s⁻¹ at 520 °C, about twice the paper’s Ru reference under its stated test. Co–Mo catalysts are established prior art, so this entry does not claim the first Co–Mo ammonia-decomposition catalyst or global composition novelty. Ex-situ XRD showed oxidized phases after air exposure and does not establish the working active phase.

Technology: Diffusion model guided by random-forest nitrogen-adsorption predictor

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Authors report Co3Mo1 catalyst performance under a defined gas-flow, temperature, support, loading and pretreatment protocol; Co75Mo25 is the corresponding generated candidate.
  • Still unknown: Exact composition/formulation prior art, long-term stability, energy efficiency and scale-up remain unresolved; this draft makes no global-priority claim.
  • AI role: A random-forest adsorption-energy predictor guides diffusion denoising; DFT and laboratory testing evaluate generated candidates.

Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC12814173/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review date: 16 Nov 2026. Review scope: Primary PMC full text opened; article metadata and candidate/experimental mapping reconciled using continuation audit dated 2026-11-16..

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

AI-guided reactor design improves a specific continuous-flow cycloaddition

Reac-Discovery combines learned geometry screening with automated experimental optimization of continuous-flow catalytic reactors. The peer-reviewed article reports a Schoen_G structured reactor with space–time yield of 803 g L−1 h−1 for a specific CO₂–epichlorohydrin cycloaddition, compared with 412 in the study’s initial data set. The comparison is study-specific and restricted to the authors’ reactor conditions; it does not establish industrial scale or net carbon removal. The article was published online on 13 October 2025; an earlier preprint described the platform and reaction, so this record attributes the 803 result to the journal version.

Technology: Neural geometry screening and machine-learning-guided experimental optimization

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: The paper reports a measured Schoen_G reactor space-time yield of 803 g L−1 h−1 for CO₂–epichlorohydrin cycloaddition, compared with 412 in the study’s initial data set. The comparison is limited to the article’s experimental conditions.
  • Still unknown: This study-specific result does not establish industrial scalability, net carbon removal or independent replication. The earlier preprint describes the platform and reaction; the 803 result is assigned to the journal article version.
  • AI role: Machine-learning optimization selected process parameters and topological descriptors for the reactor; the article reports the measured cycloaddition result for the optimized reactor.

Evidence sources: https://www.nature.com/articles/s41467-025-64127-1.pdf · https://www.nature.com/articles/s41467-025-64127-1

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review date: 17 Nov 2026. Review scope: Primary Nature Communications article metadata and publisher PDF front matter/result checked; result/date/affiliation source checked..

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

CRESt discovers a multielement formate-oxidation catalyst

CRESt combines multimodal models, knowledge-assisted Bayesian optimization and robotic synthesis and testing. Across more than 900 chemistries and 3,500 electrochemical tests, it identified an eight-element catalyst with a reported 9.3-fold improvement in cost-specific formate-oxidation performance. The metric is cost-specific, not a universal activity comparison.

Technology: CRESt multimodal model; knowledge-assisted Bayesian optimization

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: The source reports the study-specific result summarized here.
  • Still unknown: Independent replication and broader deployment or clinical benefit are not established by this source review.
  • AI role: The model combines composition, literature text embeddings and microstructural images; Bayesian optimization prioritizes experiments and robotic tests feed results back.

Evidence sources: https://www.nature.com/articles/s41586-025-09640-5

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review date: 16 Nov 2026. Review scope: Primary-source abstract and publication metadata reviewed; full text scope is specified in source_checks..

Editorial assessment: Impact 4/5 · Validation 3/5 · Novelty 4/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

ML-guided optimization improves photochemical cysteine arylation

The ChemRxiv v1 reports a two-stage optimization for nickel-catalyzed photochemical cysteine arylation: an embedding-based ranker screens categorical reagent choices, followed by Bayesian optimization of continuous conditions. The journal version reports 88% yield for both optimized conditions; the preprint reports 89% for the shorter second condition. The ranker was initially a weak yield predictor, with better separation after experimental iteration. The study expands a reaction class with earlier precedents; it does not establish invention of cysteine arylation or clinical use.

Technology: Embedding-based neural ranking; Gaussian-process Bayesian optimization

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: The two-stage workflow iteratively tests candidates and optimizes conditions; journal reports 88% yields for both conditions, while v1 reports 89% for the shorter condition.
  • Still unknown: Exact prior conditions and independent replication remain unreviewed; preprint and journal yield discrepancy retained.
  • AI role: Learned embeddings rank categorical reagent conditions; later Bayesian optimization selects continuous conditions using feedback from laboratory experiments.

Evidence sources: https://www.cambridge.org/engage/chemrxiv/article-details/68cc1a7323be8e43d6f4b2fb

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review date: 16 Nov 2026. Review scope: Primary ChemRxiv v1 record/date and version content verified in continuation audit; publisher indexed article Methods/Results freshly checked; ACS direct fetch returned 403. Journal exact online-first day remains unverified..

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

DiffSyn guides synthesis of a high-silica UFI zeolite

DiffSyn is a diffusion model trained on synthesis recipes and conditioned on zeolite structure and organic template. Researchers synthesized a UFI material using generated routes and report an ICP Si/Al ratio of 19.0. UFI is an existing framework; the result is a model-guided synthesis route and measured composition, not discovery of a new framework. Expected thermal stability is not presented as a measured outcome.

Technology: DiffSyn generative diffusion model

Publication: Preprint · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Preprint
  • Test type: direct test
  • Tested: The primary source reports the study-specific result summarized here.
  • Still unknown: Independent replication and broad deployment are not established by this review.
  • AI role: A diffusion model proposes synthesis routes from structure/template conditioning; researchers synthesize and characterize a candidate material.

Evidence sources: https://arxiv.org/abs/2509.17094

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review date: 16 Nov 2026. Review scope: Primary-source indexed article or abstract and bibliographic metadata reviewed; complete source scope is given in source_checks..

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

FlowER predicts ten-step pyrazole pathways consistent with known product selectivity

FlowER predicted two ten-step pathways to pyrazole products already reported in 2024 patents. Human researchers supplied acidic aqueous conditions omitted from the reaction database. Quantum-chemical calculations of intermediate energies favored the experimentally major product, consistent with the reported 8:2 product ratio; the kinetic interpretation assumes the Bell–Evans–Polanyi principle. This is a computational explanation of existing outcomes, with no new reaction experiment or claim of a previously unknown mechanism. Review covered the February 18, 2025 preprint results and the August 20 journal paper’s abstract, metadata and extended-data captions; the journal main text was not assessed.

Technology: FlowER (flow matching on bond-electron matrices)

Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: retrospective or benchmark
  • Tested: Specific pyrazole-pathway predictions compared with known patent products and supplemented by calculated intermediate free energies.
  • Still unknown: No new experimental confirmation; mechanistic priority not established. The kinetic argument assumes Bell–Evans–Polanyi reasoning; intermediate free energies are not measured transition-state barriers.
  • AI role: FlowER generated candidate electron-redistribution pathways; authors supplied conditions and conducted quantum-chemical analysis.

Evidence sources: https://arxiv.org/html/2502.12979v1 · https://www.nature.com/articles/s41586-025-09426-9?error=cookies_not_supported

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: in the top 1% of works of the same type, year and subfield (OpenAlex work metrics, 58 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 3/5 · Validation 2/5 · Novelty 4/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Machine learning uncovers new small-molecule chloride transporters from DrugBank

After high-throughput testing 1,894 compounds and confirming 59 chloride transporters, researchers trained and evaluated machine-learning classifiers, then prioritized DrugBank compounds for wet-lab tests. Five of ten selected drugs—bazedoxifene, centhaquine, brilliant green, tamoxifen and nevanimibe—were classified by the authors as active chloride transporters, with an estimated 50–60% hit rate under their assays.

Technology: Machine-learning binary classifiers and consensus QSAR models trained on experimentally measured chloride-transport activity

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: After high-throughput testing 1,894 compounds and confirming 59 chloride transporters, researchers trained and evaluated machine-learning classifiers, then prioritized DrugBank compounds for wet-lab tests.
  • Still unknown: The paper combines a large primary high-throughput screen with a later ML-led DrugBank virtual screen; do not attribute all 59 initial hits to ML. The five drug-repurposing hits were measured in artificial liposome assays, not cells, animals or patients. The paper discusses assay artifacts: quinacrine/brilliant green interfere with the fluorescent assay, HCl salts can affect electrode readings, and evidence for brilliant green was weaker than for the other confirmed compounds. Authors report 5/10 active among selected candidates (or 7/12 including two known active compounds), not a generalizable clinical hit rate.
  • AI role: After initial experimental discovery data were collected, machine-learning models ranked DrugBank candidates for transport activity. Researchers synthesized no new candidates for the virtual screen; instead, they tested ten prioritized available compounds in liposome chloride-transport assays. Five were classed as active; results extend these compounds' known pharmacology with membrane chloride transport activity.

Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC12368578/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: at percentile 56 of works of the same type, year and subfield (OpenAlex work metrics, 2 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 2/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

ByteDance Seed

Generative model identifies high-conductivity liquid electrolyte formulations

A predictive model and permutation-invariant generative approach were used to design liquid electrolyte mixtures. The primary v3 preprint reports three experimentally identified formulations with high ionic conductivity and anion-concentrated solvation structure. The later journal abstract additionally mentions cycling stability; that claim is excluded here because its version-specific test was not reviewed.

Technology: Predictive model; permutation-invariant mixture model; generative design

Publication: Preprint · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Preprint
  • Test type: direct test
  • Tested: The primary source reports the study-specific result summarized here.
  • Still unknown: Independent replication and commercial deployment are not established.
  • AI role: Models predict mixture properties and generate candidate formulations; researchers experimentally measure selected mixtures and solvation structure.

Evidence sources: https://arxiv.org/html/2504.18728v3

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review date: 16 Nov 2026. Review scope: Primary article/preprint text and metadata as listed in source_checks..

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Machine learning helps select nickel(I)–phenyl complexes that insert CO₂ at room temperature

The researchers calculated descriptors for 132 bidentate phosphine ligands, used k-means clustering to identify 21 candidate ligands, then applied DFT barrier calculations and human selection before synthesis. They isolated and characterized Ni(I)–phenyl complexes, including an X-ray-confirmed i-Pr-Xantphos complex. The i-Pr-Xantphos and dtbpf complexes underwent CO₂ insertion at room temperature in the reported experiments; nickel carboxylation was also tested. The study does not establish CO₂ insertion or Ni(I) chemistry generally as new; its result concerns bounded ligand selection/speciation.

Technology: Unsupervised k-means clustering of DFT-derived ligand descriptors

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Synthesis, X-ray characterization, CO₂ insertion assays and catalytic carboxylation tests reported for selected Ni complexes.
  • Still unknown: Only representative ligands from predicted groups were experimentally evaluated; prior Ni(I)-CO₂ chemistry exists; results do not imply commercial carbon capture.
  • AI role: Unsupervised clustering of computed descriptors identified a set of ligands to evaluate; DFT and researcher selection further narrowed candidates before synthesis and testing.

Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC12395408/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review date: 16 Nov 2026. Review scope: Primary JACS full text, abstract, methods/results, chronology and affiliations..

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Progressive learning screened single-atom oxide catalysts for acidic oxygen evolution

A progressive-learning workflow combining density-functional theory and machine learning screened 261 single-atom-doped metal oxides and identified nine candidates with predicted OER overpotential below 0.5 V. The publisher abstract reports experimental evaluation confirming predicted activity and durability for Mn single atoms on RuO₂ under acidic conditions. Candidate-wide values are theoretical predictions; only the specified material’s validation is summarized as experimental.

Technology: Progressive learning with DFT-derived adsorption-energy features and machine learning

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: The abstract reports experimental evaluation confirming predicted activity and durability for MnSA-RuO₂ under acidic conditions. The nine candidates and their <0.5 V overpotentials are theoretical screening outcomes, not nine experimental validations.
  • Still unknown: The abstract does not provide detailed experimental conditions or an independent comparison; those are not inferred.
  • AI role: Progressive learning combined DFT with machine learning, using adsorption-energy predictions as auxiliary features to screen single-atom-doped oxide catalysts.

Evidence sources: https://onlinelibrary.wiley.com/doi/10.1002/anie.202510965

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review date: 17 Nov 2026. Review scope: Primary publisher abstract, first-publication date, and author-affiliation metadata; full article and supporting information not reviewed..

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Skala: a deep-learning exchange-correlation functional

It comes close to the 1 kcal/mol threshold on the W4-17 benchmark (0.85 kcal/mol on the single-reference subset) and is competitive with leading hybrid functionals on GMTKN55 at lower cost. It was trained on about 150,000 high-accuracy energy differences, roughly 100 times the previous dataset size. Accuracy is shown only within the training data's chemical space; it was not yet public at the time.

Technology: Skala (neural network for DFT)

Family Skala: profile · models page

Publication: Preprint · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Preprint
  • Test type: retrospective or benchmark
  • Tested: Skala was evaluated on the W4-17 atomization benchmark, reaching 0.85 kcal/mol on the single-reference subset, and on GMTKN55, where it was competitive with hybrid functionals.
  • Still unknown: Accuracy is claimed only within the training data chemical space, expansion to broader chemistry is at an early stage, and the functional was not yet public.
  • AI role: A deep-learning exchange-correlation functional learned from about 150,000 high-accuracy energy differences.

Evidence sources: https://www.microsoft.com/en-us/research/blog/breaking-bonds-breaking-ground-advancing-the-accuracy-of-computational-chemistry-with-deep-learning/ · https://arxiv.org/abs/2506.14665

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: blog. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Editorial assessment: Impact 4/5 · Validation 2/5 · Novelty 4/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Generative machine learning nominates a chiral ligand validated in asymmetric C–H activation

A transfer-learned chemical language model generated novel chiral amino-acid ligands for palladium-catalyzed asymmetric β-C(sp³)–H activation, while an ensemble of 30 fine-tuned deep-learning regressors predicted reaction enantioselectivity. From 490 generated ligand candidates, mechanistic and synthetic filters retained 73; the authors tested selected reactions prospectively without disclosing the model-predicted enantioselectivity to the experimentalists. One highlighted reaction using generated ligand L5 gave 94% ee, matching the model prediction of 94%.

Technology: Transfer-learned ULMFiT chemical language model (FnG) for ligand generation and a 30-model deep-learning ensemble (EnP) for enantioselectivity prediction

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: From 490 generated ligand candidates, mechanistic and synthetic filters retained 73; the authors tested selected reactions prospectively without disclosing the model-predicted enantioselectivity to the experimentalists. The coauthor collaboration tested newly generated candidates; blinding experimentalists to model predictions is distinct from independent follow-up replication.
  • Still unknown: The accepted manuscript reports proof-of-concept prospective chemistry rather than a broadly validated catalyst family: only a small, expert-selected subset was tested and the authors report that some generated ligands were infeasible or failed to form product. The highlighted result is a ligand for an established C–H activation reaction, not a new reaction class; it does not establish scale-up or utility beyond tested conditions. The selected reaction gave 94% ee, but the accessible text table names it L5 and summarizes substrate/coupling partner graphically rather than in searchable prose.
  • AI role: FnG generated candidate chiral ligands from a model fine-tuned on known ligands, while EnP ranked reactions formed by pairing generated ligands with known substrates/coupling partners. The researchers selected a small set for blinded prospective wet-lab testing; the paper reports one highlighted new ligand/reaction at 94% ee, matching prediction.

Evidence sources: https://pubs.rsc.org/en/content/articlepdf/2025/sc/d5sc01098e

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: at percentile 65 of works of the same type, year and subfield (OpenAlex work metrics, 5 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 2/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Machine-learning-guided Co–Mn diatomic oxygen-reduction catalyst

A machine-learning-derived theoretical activity map guided a nitrogen-bridged Co–Mn catalyst proposal. The study reports synthesized Co-N-Mn/NC with a 0.90 V half-wave potential and 271 mW cm−2 peak power density in its alkaline oxygen-reduction and zinc–air tests. These are study-specific laboratory results, not commercial battery performance.

Technology: Machine-learning activity map; theoretical descriptor modeling

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Reported Co-N-Mn/NC catalyst was tested for ORR and in a zinc–air battery; paper reports 0.90 V and 271 mW cm−2.
  • Still unknown: Atom-specific structure attribution, training overlap, prior composition art and external replication remain open.
  • AI role: Calculated activity map informed candidate geometry and pair selection before synthesis.

Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC12134119/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review date: 16 Nov 2026. Review scope: Primary source checked for identity, publication date, AI workflow and the bounded experimental result stated..

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Active learning discovers eight experimental binders for the LRRK2 WDR domain

A free-energy molecular-dynamics and machine-learning active-learning workflow optimized compounds against the understudied LRRK2 WDR domain. Of 35 analogs experimentally tested, eight showed measurable binding confirmed by surface plasmon resonance and, for fluorinated compounds, an orthogonal 19F-NMR assay; measured dissociation constants ranged from 18 to 230 μM.

Technology: Active-learning-guided relative-binding-free-energy (AL-RBFE) workflow using molecular-dynamics thermodynamic-integration calculations and an ML affinity-prediction model

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Of 35 analogs experimentally tested, eight showed measurable binding confirmed by surface plasmon resonance and, for fluorinated compounds, an orthogonal 19F-NMR assay; measured dissociation constants ranged from 18 to 230 μM.
  • Still unknown: The experiments establish ligand binding, not functional inhibition or a therapeutic effect. Confirmed Kd values (18–230 μM) indicate modest affinity, and the work is an optimization round around two previously confirmed challenge hits rather than an initial target hit discovery. This is not evidence of Parkinson’s disease efficacy.
  • AI role: An iterative ML model selected ligand analogs for expensive free-energy simulations, using resulting calculations to choose compounds for further rounds and experimental validation. Of 35 procured/tested compounds, eight were confirmed as binding to LRRK2 WDR by SPR, with 19F-NMR used for orthogonal confirmation of fluorinated molecules.

Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC12152950/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: at percentile 79 of works of the same type, year and subfield (OpenAlex work metrics, 6 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 2/5 · Validation 3/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Machine-learning-guided screening validates 14 previously unreported drug–GPCR interactions

Chemical Space Neural Networks used network-neighborhood labels to predict drug–target interactions and were evaluated with a yeast biosensor screen of 3,773 compound–GPCR pairs (539 compounds and seven human GPCRs). The authors report 14 experimentally significant interactions with no prior literature/database evidence in the relevant chemical neighborhoods, tentatively linking some to adverse effects. This is an in-vitro receptor-screen discovery, not clinical proof of drug side effects.

Technology: Chemical Space Neural Networks (CSNN), neighborhood-to-prediction model using network homophily and graph neural networks with labels as features

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: The authors report 14 experimentally significant interactions with no prior literature/database evidence in the relevant chemical neighborhoods, tentatively linking some to adverse effects.
  • Still unknown: Yeast assay is a heterologous in-vitro system and does not establish physiological effects in humans. The article describes the 14 interactions as potentially linked to adverse effects, not confirmed clinical adverse events. Some assay limitations and false positives are discussed; precision for predicted positives is limited.
  • AI role: Used chemical-neighborhood labels to prioritize/annotate compound–GPCR interactions; model outputs were compared with high-throughput yeast biosensor measurements to identify previously unreported hits.

Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC12048553/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: at percentile 85 of works of the same type, year and subfield (OpenAlex work metrics, 8 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 2/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

MOFGen: agentic AI generates MOF linkers and five AI-designed MOFs are synthesized

MOFGen, a multi-agent system combining language models, diffusion models and quantum-mechanical checks, generated hundreds of thousands of novel MOF structures and synthesizable linkers. The authors report that five AI-designed MOFs were synthesized through high-throughput experiments. Evidence caveat: the abstract describes the validation only generally, the structure counts are computational, and this is an arXiv preprint not confirmed as peer-reviewed.

Technology: MOFGen (multi-agent LLM system with diffusion model)

Publication: Preprint · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Preprint
  • Test type: direct test
  • Tested: The abstract says five AI-designed MOFs were synthesized through high-throughput experiments; the validation is described only generally.
  • Still unknown: Generative models face challenges ensuring synthesizability; the abstract lists no other limitations.
  • AI role: A multi-agent system: a language model proposed compositions, a diffusion model generated structures, quantum-mechanical agents filtered candidates, and feasibility agents applied rules and machine learning.

Evidence sources: https://arxiv.org/abs/2504.14110

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review scope and date are not recorded; source checked does not establish full-text review.

Editorial assessment: Impact 4/5 · Validation 3/5 · Novelty 4/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Machine-learning search of archived spectra reveals vinyl–NHC coupling products in palladium catalysis

MEDUSA Search mined archived high-resolution mass spectra and uncovered previously unreported vinyl–N-heterocyclic-carbene (NHC) coupling products in Pd/NHC-catalyzed Mizoroki–Heck reactions. Researchers retrieved the corresponding reaction records and independently tested the proposed transformation with multiple palladium–NHC complexes; ultrahigh-resolution mass spectrometry and MS/MS supported the product formulas and structures. The work also found an ethyl–NHC product across tested reaction mixtures, suggesting additional catalyst-transformation pathways.

Technology: MEDUSA Search, an isotope-distribution-driven HRMS search engine with machine-learning ion-threshold regression and false-positive classification

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: MEDUSA Search mined archived high-resolution mass spectra and uncovered previously unreported vinyl–N-heterocyclic-carbene (NHC) coupling products in Pd/NHC-catalyzed Mizoroki–Heck reactions.
  • Still unknown: The claimed novelty is specific to vinyl–NHC coupling products in the studied Pd/NHC catalytic context. HRMS formulas alone cannot distinguish structural isomers, so the paper's MS/MS, ultrahigh-resolution measurements, reaction notebooks and independent experiments are essential evidence. The authors state that most other detected ions could not be associated confidently with reaction conditions because of incomplete metadata; those are not counted here.
  • AI role: Researchers generated candidate product-ion formulas from reaction fragments and catalysts; MEDUSA Search screened archived spectra using isotope-pattern matching, with two ML models estimating ion-presence thresholds and filtering likely false positives. It surfaced previously overlooked ions in old spectra; follow-up experiments with different NHC ligands and MS/MS tests checked the proposed vinyl–NHC products.

Evidence sources: https://www.nature.com/articles/s41467-025-56905-8.pdf

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: in the top 10% of works of the same type, year and subfield (OpenAlex work metrics, 28 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Machine learning screens fluoroether electrolytes for lithium-metal batteries

The study applies machine learning to a virtual library of 5,576 electrolyte combinations, then synthesizes and tests two candidate fluoroethers. The paper reports nearly 99.5% Coulombic efficiency in specified Li||NMC811 testing. This is a bounded cell result, not evidence that all virtual candidates were made or that either is commercially deployable.

Technology: Machine-learning electrolyte property and synthesizability screening

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Primary paper and PubMed abstract support virtual screening, two synthesized candidates and cell testing.
  • Still unknown: Candidate-specific assignment and exact cycling conditions behind near-99.5% metric need supplementary mapping; no commercial cell claim.
  • AI role: ML screens electrolyte property/synthesizability combinations for synthesis and Li-metal cell assessment.

Evidence sources: https://doi.org/10.1002/anie.202424493

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review date: 16 Nov 2026. Review scope: Primary source checked for identity, date, model role and bounded reported result..

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Machine learning screening identifies an efficient molecular photocatalytic CO2-reduction system

A machine-learning-accelerated descriptor workflow screened 3,444 molecular photosensitizer–catalyst systems, including 180,000 conformations, and selected six candidates for experimental validation. The primary article reports that the optimal tested system reached a turnover number of 4,390 and uses time-resolved spectroscopy and first-principles calculations to examine dipole coupling. The reported outcome is bounded to the tested system and the study’s screening scope.

Technology: Machine-learning-accelerated high-throughput screening using descriptors for photosensitization, electron transfer, and catalysis

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: The article reports experimental validation of the machine-learning-screened molecular photocatalytic systems; its optimal system achieved a turnover number of 4,390. The study screened 3,444 systems and selected six promising candidates.
  • Still unknown: The summary does not identify a catalyst or photosensitizer structure because the accessible primary indexed passage did not unambiguously tie a named complex to the reported maximum. Performance is specific to the reported experimental conditions; deployment and independent replication are not established.
  • AI role: Machine learning accelerates high-throughput screening with descriptors spanning photosensitization, electron transfer and catalysis, reducing the 3,444-system search to six candidates selected for experimental validation.

Evidence sources: https://www.nature.com/articles/s41929-025-01291-z · https://chem.ahnu.edu.cn/info/1028/28946.htm

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review date: 17 Nov 2026. Review scope: Primary Nature Catalysis indexed article text, abstract, extended-data descriptions, metadata and affiliation block; institutional research announcement checked as corroboration..

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Org-Mol prioritizes two experimental immersion-coolant candidates

A 3D transformer pretrained on molecular structures and fine-tuned for physical-property prediction screened millions of ester candidates. Researchers synthesized and characterized two selected candidates. One did not meet the preferred thermal-conductivity threshold. This is experimental candidate evaluation, not industrial deployment or environmental-safety validation.

Technology: Org-Mol 3D molecular transformer

Publication: Preprint · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Preprint
  • Test type: direct test
  • Tested: The primary source reports the study-specific result summarized here.
  • Still unknown: Independent replication and broad deployment are not established by this review.
  • AI role: A pretrained molecular transformer predicts bulk properties to prioritize candidate esters; researchers assess feasibility, synthesize and measure two candidates.

Evidence sources: https://arxiv.org/abs/2501.09896

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review date: 16 Nov 2026. Review scope: Primary-source indexed article or abstract and bibliographic metadata reviewed; complete source scope is given in source_checks..

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Neural network guides a Ru–Ti–Mn oxygen-evolution electrode

A backpropagation neural network fitted composition/activity and deactivation data and suggested a Ru:Ti:Mn ratio of 0.26:0.26:0.48. The synthesized electrode had measured composition near 0.24:0.28:0.48 and reported 165.2 mV overpotential at 10 mA cm−2 in the acidic assay. Model optimum and measured composition are distinct; reported long-duration tests apply only under the study’s specific conditions.

Technology: Backpropagation neural network

Publication: Preprint · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Preprint
  • Test type: direct test
  • Tested: Research Square v1 reports measured Ru/Ti/Mn electrode and acidic OER result.
  • Still unknown: Exact prior composition overlap and whether candidate was included in training data remain open.
  • AI role: Neural network fit experimental composition/activity/durability data to suggest a candidate composition, which was synthesized and tested.

Evidence sources: https://doi.org/10.21203/rs.3.rs-5208867/v1

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review date: 16 Nov 2026. Review scope: Primary source checked for identity, publication date, AI workflow and the bounded experimental result stated..

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Machine learning screened metal oxides for acetylene semihydrogenation

A neural-network model was used to predict metal-oxide catalyst performance for acetylene semi-hydrogenation. The authors report experimental validation of predictions for CuO, ZnO and V2O5, and report 99.6% conversion with 90% ethylene selectivity for CuO under the abstract-described conditions. The entry is limited to the publisher abstract; detailed model validation, candidate selection and full assay context were not reviewed.

Technology: Neural-network prediction of metal-oxide catalyst performance

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: The publisher abstract says CuO, ZnO and V2O5 predictions were experimentally validated; it reports 99.6% conversion and 90% ethylene selectivity for CuO at 50–100°C.
  • Still unknown: Detailed selection chronology, model split/performance, full test conditions and results for ZnO and V2O5 were not established from the reviewed abstract. The date is the publisher-displayed date, not a claim of first public appearance.
  • AI role: A neural-network model predicted catalyst performance and informed the metal-oxide candidates reported as experimentally validated.

Evidence sources: https://www.sciencedirect.com/science/article/pii/S0926860X24004149 · https://homepage.zjut.edu.cn/ljh_en/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review date: 17 Nov 2026. Review scope: Primary publisher abstract for identity, displayed date, model and reported experiment; official Zhejiang University of Technology faculty profile/publication list for named contributor affiliation. Full paper methods and supplement not reviewed..

Editorial assessment: Impact 2/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

GPT-4 agents guide chemical synthesis development and purification

The November 2024 Nature Communications paper describes six GPT-4 agents guiding copper/TEMPO alcohol oxidation, from literature search through optimization and purification. An agent-assisted scale-up produced 915 mg of product at 86% isolated yield and above 98% purity. Additional reaction tasks were demonstrated. The workflow involved chemists: thin-layer chromatography was performed manually, so this was not fully autonomous synthesis.

Technology: LLM-RDF (six GPT-4 agents)

Family GPT-4: profile · models page

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Physical synthesis, optimization, scale-up and product purification.
  • Still unknown: Not assessed: supplements; manual TLC and chemist participation are explicit in the paper.
  • AI role: Six GPT-4 agents supported literature, design, execution, analysis, separation and interpretation.

Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC11585555/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: in the top 1% of works of the same type, year and subfield (OpenAlex work metrics, 189 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 4/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

ClickGen designs PARP1 inhibitors with measured nanomolar activity

ClickGen combines reaction-based molecular generation, reinforcement learning and inpainting to generate synthesizable candidate structures. In a PARP1 case study, researchers selected and synthesized three generated lead molecules; two showed nanomolar PARP1 inhibition in vitro, with reported IC50 values of 19.24 ± 1.63 nM and 69.73 ± 14.57 nM. The two compounds also showed antiproliferative activity in several tested cancer cell lines and lower toxicity in the reported normal-cell assays than the most toxic lead compounds.

Technology: ClickGen deep-learning reaction-based molecular generator with reinforcement learning, Monte Carlo tree search and an inpainting generator

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: In a PARP1 case study, researchers selected and synthesized three generated lead molecules; two showed nanomolar PARP1 inhibition in vitro, with reported IC50 values of 19.24 ± 1.63 nM and 69.73 ± 14.57 nM.
  • Still unknown: Three selected molecules were synthesized and tested in vitro; the article reports two with nanomolar PARP1 inhibition. Cell-line antiproliferative and toxicity assays do not show tumor efficacy or safety in animals or people. The reported IC50 values are from triplicate measurements in the study. This record captures the PARP1 experimental result, not the paper’s model benchmark comparisons.
  • AI role: The system generated candidate molecules by assembling reaction-compatible synthons and using reinforcement learning/inpainting to guide novelty and binding-oriented generation. Researchers computationally filtered and selected compounds, synthesized three leads and tested them; PARP1 inhibition and cell effects were experimentally measured, not generated by the model.

Evidence sources: https://www.nature.com/articles/s41467-024-54456-y.pdf

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: in the top 10% of works of the same type, year and subfield (OpenAlex work metrics, 45 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 2/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Machine learning selects silver-electrode CO₂ electrolysis conditions

A model trained on 56 experimental observations selected electrode-fabrication and electrolysis conditions for silver-catalyzed CO₂ reduction. Researchers experimentally validated the proposed conditions against a desired hydrogen-to-carbon-monoxide ratio and CO efficiency. This is an assay-specific process-optimization result for a known silver catalyst, not a newly discovered catalyst or evidence of industrial carbon-neutral operation.

Technology: Random-forest regression; kernel ridge regression

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: The paper reports 56 experimental observations and experimental validation of selected conditions for the target product ratio and CO faradaic efficiency.
  • Still unknown: Independent replication and performance outside stated assay conditions are not established.
  • AI role: ML models trained on experimental outcomes estimate selectivity across fabrication and electrolysis parameters; inverse search proposes conditions for follow-up testing.

Evidence sources: https://d-nb.info/1349018996/34

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review date: 16 Nov 2026. Review scope: Primary author manuscript PDF read through abstract, results and conditions; exact online date from source ledger, issue metadata says 2024..

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

OCx24: experimental catalysts for evaluating prediction models

A dataset from Meta FAIR, the University of Toronto, and VSParticle covers over 600 mixed-metal catalysts made from 13 elements and tested under the same conditions, alongside analysis of over 19,000 materials. Hydrogen-evolution predictions flagged platinum-containing alloys as promising, although platinum was absent from training data. Fewer than 25% of produced catalysts matched the target; this was a pilot.

Technology: Catalyst prediction models (Open Catalyst)

Publication: Preprint · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Preprint
  • Test type: direct test
  • Tested: Computational predictions were compared with lab results: 441 gas diffusion electrodes tested in zero-gap electrolysis and 572 characterized samples, alongside DFT-verified adsorption energies on about 20,000 materials.
  • Still unknown: Platinum was ranked as a top hydrogen-evolution candidate but was not experimentally tested, and the abstract lists no explicit limitations.
  • AI role: AI-accelerated relaxations (685 million) fed a data-driven Sabatier volcano that flagged platinum-containing alloys as promising.

Evidence sources: https://arxiv.org/abs/2411.11783

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: blog. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Prospective deep-learning screen identifies nanomolar IRAK1 inhibitor scaffolds

Ro5's prospective HydraScreen structure-based deep-learning screen ranked compounds for IRAK1 and was evaluated against a 47,000-compound robotic high-throughput screen. The study reports that 23.8% of IRAK1 hits were in HydraScreen's top 1% and identifies three nanomolar scaffolds, two described in the abstract as novel IRAK1 candidates.

Technology: HydraScreen structure-based deep-learning virtual screening, with SpectraView knowledge-graph target evaluation

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: The study reports that 23.8% of IRAK1 hits were in HydraScreen's top 1% and identifies three nanomolar scaffolds, two described in the abstract as novel IRAK1 candidates. The new candidate experiments were conducted within the Ro5/Strateos discovery collaboration.
  • Still unknown: This is prospective early-stage hit identification, not drug development or disease efficacy. The assay validated kinase inhibition in vitro. Authors are affiliated with the companies that developed the tools. The abstract describes two scaffolds as novel candidates, while detailed similarity analysis indicates one of three nanomolar scaffolds had a similar known active; do not call all hits unprecedented.
  • AI role: SpectraView selected IRAK1 as the study target; HydraScreen scored docked ligand-pose ensembles to rank compounds before prospective screening. Researchers compared rankings with automated high-throughput assay results from a 47,000-compound library and reported nanomolar IRAK1 hit scaffolds, two described as novel candidates.

Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC11566907/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: at percentile 77 of works of the same type, year and subfield (OpenAlex work metrics, 9 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 2/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Machine-learning-guided optimization finds low-loading conditions for platinum-catalyzed amide reduction

A machine-learning-assisted reaction-optimization study found conditions for reducing N,N-dimethylacetamide to dimethylethylamine using the known platinum complex [Pt(DMS)₂Cl₂]. The selected condition used 0.01 mol% catalyst and reached complete conversion in two hours at 39 °C without solvent, with a reported turnover frequency of 5,002 h⁻¹. The chemical catalyst identity is not claimed as new.

Technology: Sunthetics machine-learning reaction-optimization platform

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: The selected condition used 0.01 mol% catalyst and reached complete conversion in two hours at 39 °C without solvent, with a reported turnover frequency of 5,002 h⁻¹.
  • Still unknown: This is a single-substrate reaction-condition optimization on N,N-dimethylacetamide. [Pt(DMS)₂Cl₂] is a known complex; the reported finding is the ML-assisted optimized condition and measured performance, not discovery of a new platinum compound. The best-performing condition is reported after human-guided interpretation as well as model recommendations. The paper does not establish transferability to other amides, industrial scale-up or general performance across catalyst classes.
  • AI role: The ML platform recommended reaction conditions across iterative experiments, using observed reaction outcomes to refine proposed conditions and reduce the number of trials. The team combined those recommendations with chemist interpretation and tested the predicted conditions experimentally; the source reports the optimized platinum-catalyzed reduction as a study-specific result.

Evidence sources: https://pubs.rsc.org/en/content/articlehtml/2024/cc/d4cc05273k

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: at percentile 59 of works of the same type, year and subfield (OpenAlex work metrics, 7 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 2/5 · Validation 3/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Mobile autonomous robots for exploratory synthetic chemistry

Mobile robots move samples between a synthesis platform, a UPLC-MS spectrometer, and a benchtop NMR, while a heuristic decision-maker evaluates data and chooses reactions. Demonstrated on structural diversification, supramolecular chemistry, and photochemical synthesis. The AI is heuristic and there is no speed or accuracy metric; the announcement says robots will not have a ‘Eureka’ moment.

Technology: Heuristic decision-making with mobile robots

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Mobile robots were demonstrated on three exploratory chemistry problems (structural diversification, supramolecular chemistry, photochemical synthesis); the page reports no reaction counts or accuracy metrics, only that the AI made the same or similar decisions as a human researcher.
  • Still unknown: The page reports no quantitative accuracy metrics, only two robots were used, and the system lacks a 'Eureka' moment.
  • AI role: A heuristic decision-maker processes analytical data and autonomously decides whether a reaction proceeds to the next step.

Evidence sources: https://news.liverpool.ac.uk/?p=123346

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: press release. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: in the top 1% of works of the same type, year and subfield (OpenAlex work metrics, 307 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 4/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Minerva identifies nickel-catalyzed Suzuki reaction conditions

Minerva combines Gaussian-process optimization with automated high-throughput experiments to search reaction conditions. In the nickel-catalyzed Suzuki campaign, the preprint reports a condition with 76 area-percent yield and 92% selectivity, while two chemist-designed screening plates did not find successful conditions. Area percent is a chromatographic assay measure, not isolated yield. The preprint and expanded journal report are one study; this does not establish invention of Suzuki coupling or clinical utility.

Technology: Minerva Gaussian-process surrogate; Bayesian acquisition; automated HTE

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Experimental 96-well optimization reports 76 AP yield and 92% selectivity; two expert-designed screening plates did not find successful conditions.
  • Still unknown: Product assay area-percent is not isolated yield; pharmaceutical structures and independent replication remain unreviewed.
  • AI role: A GP surrogate and Bayesian acquisition choose batches for automated experiments and use measured outcomes to update condition search.

Evidence sources: https://chemrxiv.org/engage/api-gateway/chemrxiv/assets/orp/resource/item/66f976b5cec5d6c142981e4e/original/highly-parallel-optimisation-of-nickel-catalysed-suzuki-reactions-through-automation-and-machine-intelligence.pdf

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review date: 16 Nov 2026. Review scope: Primary ChemRxiv v1 full-text PDF and primary PubMed/journal metadata and abstract checked; journal full PMC direct open CAPTCHA challenged. Exact first-version content for later API case studies is excluded from draft..

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Solvent mixtures for flow batteries found through an AI–experiment cycle

PNNL and Argonne identified solvent mixtures that dissolve three times more than a compound proposed for flow batteries, exploring less than 10% of about 2,000 combinations. The model receives experimental data and guides subsequent experiments; the result concerns solubility, without demonstrating performance of a complete battery.

Technology: Machine learning in a loop with high-throughput experiments

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Active learning guided solubility measurements of 2,1,3-benzothiadiazole in binary solvents: 218 measurements from 2,101 candidates found 18 mixtures above 6.20 M; best 6.50 M. On a 98-solvent set, Bayesian optimization needed 17 ± 11 trials versus about 50 ± 27 random.
  • Still unknown: Tested one solute, binary solvents only; solubility is not the only property and viscosity, conductivity and stability were not optimized. Model accuracy on 40 test solvents was R2 = 0.63. Real battery performance not reported.
  • AI role: A Gaussian process model with an expected-improvement rule chose which solvents to measure; a robotic platform prepared samples and NMR measured solubility; humans selected features and candidate lists.

Evidence sources: https://doi.org/10.1038/s41467-024-47070-5

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: press release. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: in the top 1% of works of the same type, year and subfield (OpenAlex work metrics, 102 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Descriptor-guided Co–Co and Co–Ir oxygen electrocatalysts

Physically informed feature selection derived an ARSC descriptor from calculated catalyst properties to guide oxygen-reaction candidate design. The study reports synthesis and alkaline testing of two carbon-supported samples: ORR half-wave potentials of 0.941 and 0.937 V and OER overpotentials of 330 and 340 mV at 10 mA cm−2. These are laboratory measurements; structural analyses support but do not prove every proposed active-site configuration.

Technology: ARSC descriptor; feature selection

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Publisher PDF reports synthesized Co–Co and Co–Ir samples with stated ORR and OER tests.
  • Still unknown: Prior composition overlap and independent replication remain open; specific active-site configuration is not directly resolved for every site.
  • AI role: Calculated dual-atom catalyst descriptors selected candidates followed by synthesis and electrochemical testing.

Evidence sources: https://www.nature.com/articles/s41467-024-52519-8

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review date: 16 Nov 2026. Review scope: Primary source checked for identity, publication date, AI workflow and the bounded experimental result stated..

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Chemical language models produce experimentally confirmed dual ligands for three target pairs

A generative chemical language model was fine-tuned on known ligands for paired targets. Prospective synthesis and in-vitro assays found seven of twelve tested designs modulated both intended targets, with confirmed dual activity across FXR/sEH, FXR/THRβ and PPARδ/sEH. Activities varied substantially and were often unbalanced; the strongest reported sEH inhibitor, compound 7, had IC50 17 nM with micromolar PPARδ agonism.

Technology: Target-pair-fine-tuned chemical language model (four-layer LSTM recurrent neural network trained on SMILES, pretrained on 365,000 ChEMBL molecules)

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Prospective synthesis and in-vitro assays found seven of twelve tested designs modulated both intended targets, with confirmed dual activity across FXR/sEH, FXR/THRβ and PPARδ/sEH.
  • Still unknown: This is in-vitro activity of a small set of synthesized molecules, not evidence of therapeutic efficacy or synergistic clinical benefit. Potencies were often unbalanced; several selected candidates were active on only one intended target, and the authors caution the success rate may not generalize beyond target pairs with compatible ligand chemistry. Lead optimization remains necessary.
  • AI role: The pooled fine-tuning strategy combined known ligands of each target pair to bias de-novo SMILES generation toward both ligand classes. Researchers used model sampling frequency, similarity and docking to select molecules for synthesis; assays then tested receptor agonism and soluble epoxide hydrolase inhibition. The model proposed chemical candidates, while biological activity was measured experimentally.

Evidence sources: https://www.nature.com/articles/s41467-024-52060-8

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: in the top 10% of works of the same type, year and subfield (OpenAlex work metrics, 62 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Human-guided active learning improves nickel-cobalt methanation tests

A human-in-the-loop model selected nickel-cobalt catalysts and reaction settings for a 48-test laboratory campaign. Further selected experiments improved methane space-time yield by nearly 50% over the previous maximum in that dataset. Extrapolation remained harder than interpolation. This is a bounded catalyst-and-condition result, not evidence of industrial deployment or lifecycle climate benefit.

Technology: Gaussian-process active learning with batch acquisition

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Selected catalysts and conditions were experimentally tested; model-suggested activity trends received follow-up.
  • Still unknown: Generality, industrial durability and lifecycle carbon benefit were not established.
  • AI role: Gaussian-process active learning with batch acquisition supported the stated study; experimental and interpretive work remained with researchers.

Evidence sources: https://pubs.rsc.org/en/content/articlehtml/2024/cy/d4cy00873a

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review date: 16 Nov 2026. Review scope: Indexed primary publisher abstract, model Methods, selected Results, affiliations and first-online date; supplementary material not reviewed..

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

AI-accelerated virtual screening identifies small-molecule binders to KLHDC2 and Nav1.7

The OpenVS platform combined target-specific neural-network active learning with RosettaVS physics-based docking to search billion-scale compound libraries. The study reported seven low-micromolar KLHDC2 degron-pocket binders and four Naᵥ1.7 VSD4 binders. KLHDC2 hit C29 competed with a degron peptide at IC50 about 2.9 μM and its predicted pose was supported by a 2.0 Å crystal structure; the strongest Naᵥ1.7 hit, Z8739902234, had IC50 1.3 μM in whole-cell electrophysiology.

Technology: OpenVS active-learning platform with target-specific feed-forward neural-network triage, coupled to RosettaVS physics-based docking and RosettaGenFF-VS scoring

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: The study reported seven low-micromolar KLHDC2 degron-pocket binders and four Naᵥ1.7 VSD4 binders.
  • Still unknown: Early ligand-hit discovery only: binding/inhibition was measured in biochemical competition, biolayer interferometry and cell electrophysiology assays, not in animals or patients. The paper states further optimization is needed to improve KLHDC2 compounds toward nanomolar potency. KLHDC2 C29’s crystal structure is experimental support for one predicted binding pose; target modulation, therapeutic efficacy and safety were not established. The reported target hit rates apply to manually filtered, synthesized subsets, not the full billion-molecule library.
  • AI role: The target-specific neural network used docking-labeled data to guide which molecules were prioritized for additional expensive docking over iterative multi-billion-compound screens; RosettaVS/RosettaGenFF-VS performed physics-based docking and pose/affinity scoring. The AI-guided screen nominated molecules that were synthesized and experimentally tested. Experiments confirmed KLHDC2 degron displacement and a crystal structure matching the predicted pose, plus Naᵥ1.7 binding/inhibition in patch clamp. The paper attributes discovery to the combined AI-accelerated and physics-based workflow, not to a generative model alone.

Evidence sources: https://www.nature.com/articles/s41467-024-52061-7.pdf

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: in the top 1% of works of the same type, year and subfield (OpenAlex work metrics, 290 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

MitoReID identifies epicatechin as a cyclooxygenase-2 inhibitor

A deep-learning system called MitoReID classified drug mechanisms from time-resolved mitochondrial morphology and membrane-potential images. Among its natural-compound predictions, it nominated epicatechin as a COX-2 inhibitor; the authors report in-vitro follow-up including COX-2 activity, cellular thermal-shift and surface-plasmon-resonance assays. This is a target-mechanism finding, not evidence of clinical efficacy.

Technology: MitoReID, deep-learning re-identification model using an Inflated 3D ResNet backbone on temporal mitochondrial images

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Among its natural-compound predictions, it nominated epicatechin as a COX-2 inhibitor; the authors report in-vitro follow-up including COX-2 activity, cellular thermal-shift and surface-plasmon-resonance assays.
  • Still unknown: Evidence is biochemical/in-vitro and does not establish therapeutic efficacy or safety. MitoReID's six unseen-drug mechanism predictions are a separate task; the epicatechin COX-2 result is the specific new mechanistic finding recorded here. Full publisher page was cookie-blocked, but the primary PubMed record and publisher-indexed article text/extended-data caption were read.
  • AI role: Ranked mechanisms of action for compounds by matching learned time-dependent mitochondrial phenotypes against reference drugs; the model nominated COX-2 inhibition for epicatechin, which was tested in vitro.

Evidence sources: https://pubmed.ncbi.nlm.nih.gov/39169261/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: in the top 10% of works of the same type, year and subfield (OpenAlex work metrics, 63 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 2/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

FRASE-bot screening identifies a small-molecule ligand that selectively kills CIB1-dependent cancer cells

FRASE-bot used a neural-network nativeness score to triage protein–ligand fragments, then used the retained structural features to seed virtual screening for Calcium and Integrin Binding Protein 1 (CIB1). Of 56 purchased virtual-screen hits, 29 showed dose-dependent CIB1 displacement; compound UNC10245380 (UNC5380) had a TR-FRET displacement IC50 of 8.0 ± 2.5 μM and selectively killed CIB1-depletion-sensitive triple-negative breast cancer cells in the reported assays, with no detected killing in an insensitive line.

Technology: FRASE-bot: fragment-in-structural-environment screening with a neural-network nativeness scoring model

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Of 56 purchased virtual-screen hits, 29 showed dose-dependent CIB1 displacement; compound UNC10245380 (UNC5380) had a TR-FRET displacement IC50 of 8.0 ± 2.5 μM and selectively killed CIB1-depletion-sensitive triple-negative breast cancer cells in the reported assays, with no detected killing in an insensitive line.
  • Still unknown: The authors describe UNC10245380 as a starting chemical probe/lead, not a drug. Binding was inferred from peptide displacement in TR-FRET and a docking pose; the pose is not a solved small-molecule complex. Selective cell killing was in vitro at 30 μM, with no animal or clinical evidence. The earlier Research Square manuscript is the same underlying paper/result and is not counted separately.
  • AI role: The neural network triaged fragments by predicted native-ligand pose likelihood; selected fragments informed pharmacophore construction and virtual screening. The resulting purchased compounds were then tested in TR-FRET binding-displacement and cell-death assays. The model did not perform the wet-lab validation.

Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC11219766/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: at percentile 86 of works of the same type, year and subfield (OpenAlex work metrics, 14 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 2/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Bayesian search identifies organic metallophotocatalyst formulations

Gaussian-process models and chemist decisions guided synthesis and testing of organic photocatalysts for a specified cross-coupling reaction. A catalyst search reached 67% yield before reaction-condition optimization. This establishes a bounded formulation result for the studied reaction, not discovery of the reaction or general superiority to iridium catalysts.

Technology: Gaussian-process Bayesian optimization

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Candidate catalysts were synthesized and reaction yields measured in repeated experiments.
  • Still unknown: General substrate scope, independent replication and exhaustive molecular prior art were not assessed.
  • AI role: Gaussian-process Bayesian optimization supported the stated study; experimental and interpretive work remained with researchers.

Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC11321994/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review date: 16 Nov 2026. Review scope: Primary article abstract, selected Results, computational Methods and publication metadata; supplements not reviewed..

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

ML-guided mechanosynthesis yields a lead-free photocatalyst

Machine-learning-guided synthesis and testing identified a lead-free metal-halide perovskite for oxidative alkene cleavage. The publisher reports an 80% benzoic-acid yield under the study’s conditions, compared with 6% for the initial Cs2AgBiBr6 benchmark. This is a formulation- and reaction-specific laboratory result, not universal photocatalytic performance or industrial deployment.

Technology: Machine-learning-assisted materials selection and iterative synthesis

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Publisher article reports an experimentally tested ML-selected perovskite yielding 80% benzoic acid versus 6% for the initial benchmark in the specified reaction.
  • Still unknown: Independent replication and broader application are not established by this bounded review.
  • AI role: ML ranks perovskite candidates using experimental feedback across synthesis rounds; researchers synthesize and test the selected photocatalysts.

Evidence sources: https://advanced.onlinelibrary.wiley.com/doi/abs/10.1002/advs.202309714

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review date: 16 Nov 2026. Review scope: Publisher indexed primary full text and metadata; specific formulation and selected result reviewed. Earlier disclosure not established in this pass..

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Machine learning guided multi-metric optimization of photocatalytic CO2 reduction

Bayesian optimization guided experiments on a five-component self-assembled photocatalytic micelle system for CO₂-to-CO reduction. The authors report that optimized conditions increased measured CO yield from 1.48 ± 0.24 to 5.72 ± 0.12 μmol h⁻¹, with improved turnover and quantum-yield metrics while maintaining CO selectivity. The article’s narrative fold-change (4.2×) is inconsistent with the displayed means (about 3.9×), so this record gives the means and omits the fold claim. The findings apply to the tested aqueous micelle system and conditions.

Technology: Bayesian optimization for experimental selection; random-forest models and SHAP for later analysis

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: The article reports CO yield means of 5.72±0.12 and 1.48±0.24 μmol h−1 for optimized vs initial conditions, along with turnover and quantum-yield metrics and maintained selectivity. The narrative states a 4.2-fold yield improvement, inconsistent with these displayed means; the record does not repeat the fold value.
  • Still unknown: This demonstrates a study-specific photocatalytic system optimization, not industrially scaled carbon conversion or independent replication. The source’s stated yield fold change conflicts with its displayed means.
  • AI role: Bayesian optimization guided the experimental parameter search and selected test conditions in a five-component micelle system; random-forest models and SHAP then analyzed parameter effects.

Evidence sources: https://api.repository.cam.ac.uk/server/api/core/bitstreams/9bb1fb68-8d73-4b07-b47b-376146e130f8/content · https://pubs.acs.org/doi/pdf/10.1021/jacs.4c01305?download=true

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review date: 17 Nov 2026. Review scope: Primary ACS publisher metadata and University of Cambridge repository copy of the article full text checked for article identity, affiliations, date, optimization method, results and source discrepancy..

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

ChemCrow: a language-model agent with chemistry tools

An agent with 18 tools and GPT-4 planned and carried out the synthesis of DEET and three thiourea organocatalysts on a cloud platform, and guided discovery of a chromophore absorbing at 336 nm against a 369 nm target; evaluated on 14 cases by four expert chemists. Hallucinations, limited tasks and dual-use risks; 12 of 18 tools were released.

Technology: ChemCrow (GPT-4 with tools)

Family GPT-4: profile · models page

Publication: Preprint · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Preprint
  • Test type: direct test
  • Tested: Expert and LLM judgments were used to evaluate the agent, which planned and executed syntheses of an insect repellent and three organocatalysts and guided discovery of a chromophore.
  • Still unknown: The abstract says LLMs struggle with chemistry and lack external knowledge, and GPT-4 as evaluator could not distinguish clearly wrong completions from ChemCrow's performance.
  • AI role: An LLM agent with 18 expert-designed chemistry tools planned and executed syntheses and guided molecular discovery; the page does not name the underlying model.

Evidence sources: https://arxiv.org/abs/2304.05376

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: in the top 1% of works of the same type, year and subfield (OpenAlex work metrics, 1741 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

LAMBDAZERO generates an experimentally active soluble epoxide hydrolase inhibitor scaffold

The LAMBDAZERO generative active-learning workflow searched for synthesizable soluble epoxide hydrolase (sEH/EPHX2) inhibitors and produced a quinazoline scaffold absent from the authors' comparison set of known sEH inhibitors. A synthesized analog, UM0152893, inhibited sEH in vitro with sub-micromolar potency.

Technology: LAMBDAZERO generative active learning with deep reinforcement learning and docking

Publication: Preprint · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Preprint
  • Test type: direct test
  • Tested: The LAMBDAZERO generative active-learning workflow searched for synthesizable soluble epoxide hydrolase (sEH/EPHX2) inhibitors and produced a quinazoline scaffold absent from the authors' comparison set of known sEH inhibitors.
  • Still unknown: This is a workshop paper/preprint rather than a clinical study. The reported lead is a biochemical inhibitor of purified sEH; no cellular, animal, safety or therapeutic outcome is established. The authors report 24/35 synthesized analogs active in vitro with potency spanning roughly 90 μM to 0.4 μM; one best lead is sub-micromolar.
  • AI role: The generative model searched molecular space using docking, synthesizability and drug-likeness constraints and proposed the scaffold; researchers synthesized analogs and measured enzyme inhibition in biochemical assays.

Evidence sources: https://arxiv.org/abs/2405.01616

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Editorial assessment: Impact 2/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Iterative machine learning identifies potent inhibitors of alpha-synuclein secondary nucleation

A structure-based iterative-learning workflow combined prior α-synuclein fibril-binding hits, a junction-tree variational autoencoder, a random-forest regressor and Gaussian-process residual model to rank compounds for repeated testing. Prospective iterations produced multiple molecules inhibiting seeded α-synuclein aggregation in vitro; lead I4.05 had a reported kinetic inhibitory constant (KIC50) of 0.52 μM (95% CI 0.45–0.59 μM), compared with extrapolated values of 18.2 μM for its parent and 36.4 μM for Anle-138b in the authors’ assay. Leads also showed activity in lipid-induced and spontaneous aggregation assays.

Technology: Iterative structure-based learning workflow using a junction-tree variational autoencoder, random-forest regressor and Gaussian-process residual model

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: A structure-based iterative-learning workflow combined prior α-synuclein fibril-binding hits, a junction-tree variational autoencoder, a random-forest regressor and Gaussian-process residual model to rank compounds for repeated testing.
  • Still unknown: Evidence is biochemical/in-vitro inhibition of α-synuclein aggregation, including seeded, lipid-induced and spontaneous assays. The paper does not establish animal efficacy, neuronal protection or clinical benefit. The KIC50 and comparison values are assay-specific; no random-molecule comparator was tested, and model QSAR R² was described as modest. This is related to, but not the same paper/result as, the earlier docking-screen study that supplied starting compounds.
  • AI role: The machine-learning model ranked a ZINC-derived evaluation set for potency using earlier assay measurements, then researchers tested and fed results from successive rounds back into the model. It proposed compounds for experimental testing; aggregation assays measured outcomes. Initial fibril binders came from an earlier docking-and-screening study, so this paper reports iterative optimization rather than an entirely independent start.

Evidence sources: https://www.nature.com/articles/s41589-024-01580-x.pdf

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: in the top 10% of works of the same type, year and subfield (OpenAlex work metrics, 60 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Feature-fusion model prioritizes antibacterial Ugi products

LSCNN ranked a combinatorial Ugi library after model development on focused activity data. The paper reports six active products among ten selected combinations against MRSA; purified H4–H6 had MIC and MBC of 12 μM. Different screening phases and cutoffs mean these figures are not a randomized general hit-rate comparison.

Technology: LSCNN feature-fusion neural network

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Primary article reports 6/10 selected combinations active against MRSA and purified H4–H6 MIC/MBC results.
  • Still unknown: Prior chemical-structure overlap, full independent replication, and clinical efficacy are not established.
  • AI role: Model ranks products from combinatorial Ugi space for synthesis and testing.

Evidence sources: https://pubs.rsc.org/en/content/articlepdf/2024/sc/d3sc06441g · https://pmc.ncbi.nlm.nih.gov/articles/PMC11041243/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review date: 16 Nov 2026. Review scope: Primary source checked for identity, date, model role and bounded reported result..

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Machine learning screening reveals ABBV-744 as a modulator of P2Y purinergic receptors

Researchers used consensus machine-learning classifiers trained on known P2Y6 agonists to select compounds for pharmacological testing. The anticancer drug ABBV-744 showed non-surmountable inhibition of UDP-induced P2Y6 receptor activation and enhanced P2Y1 receptor activation in cellular assays, revealing previously unreported purinergic-receptor activity for this molecule.

Technology: Consensus of deep learning, AdaBoost, naive Bayes, k-nearest neighbors, logistic regression, random forest, SVM and XGBoost classifiers

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Researchers used consensus machine-learning classifiers trained on known P2Y6 agonists to select compounds for pharmacological testing.
  • Still unknown: The paper found no novel P2Y6 agonists, which was the initial aim. ABBV-744 inhibited P2Y6 activation at relatively high micromolar concentration (reported IC50 about 75.7 µM); its precise inhibitory mechanism was not determined and requires further assays. P2Y1 enhancement and weak P2Y14 binding effects are assay-specific observations, not clinical pharmacology or therapeutic benefit.
  • AI role: Models trained on curated P2Y6 agonist activity data prioritized molecules from compound libraries for experimental screening. Functional calcium-transient assays in receptor-expressing cells and fluorescent binding assays were used to test predictions; ABBV-744 was the principal confirmed hit with unanticipated activity across P2Y receptor subtypes.

Evidence sources: https://link.springer.com/article/10.1007/s11302-024-10003-4

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: at percentile 88 of works of the same type, year and subfield (OpenAlex work metrics, 10 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 2/5 · Validation 3/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Machine learning selected nucleoside derivatives for hydrogel testing

A machine-learning model trained on 71 previously reported nucleoside derivatives was applied to an external set, from which the authors selected 24 candidates for hydrogel testing. They report gel formation for 20 of the 24 tested molecules and characterize cation-independent hydrogel behavior for 8-aminoguanosine and 8-hydroxyguanosine. A 2023 Sichuan University patent by the same team had already disclosed the model, 12 selected candidates and 10 gelation positives, including these two derivatives. This entry represents one study/campaign and does not claim that the 2024 article was the first public disclosure or that the compounds were novel.

Technology: Machine-learning classifier using molecular descriptors to predict nucleoside derivative hydrogel formation

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: The paper reports that 20 of 24 externally selected nucleoside derivatives formed hydrogels in experimental validation. It also reports cation-independent hydrogel behavior for 8-aminoguanosine and 8-hydroxyguanosine.
  • Still unknown: A 2023 patent application from the same university/team already disclosed the same model, 12 selected derivatives, 10 gelation positives, and included 8-AG and 8-OHG. Do not treat the overlapping paper and patent as independent discoveries or claim first disclosure/compound novelty. Hydrogel behavior does not establish clinical benefit.
  • AI role: The ML model predicted gel-forming ability and selected nucleoside derivatives for external experimental validation.

Evidence sources: https://www.nature.com/articles/s41467-024-46866-9 · https://pmc.ncbi.nlm.nih.gov/articles/PMC10960799/ · https://pubmed.ncbi.nlm.nih.gov/38521777/ · https://patents.google.com/patent/CN116741306A/en

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review date: 17 Nov 2026. Review scope: Primary Nature article page and PubMed Central indexed full text for model selection and experimental outcomes; PubMed record for date and affiliations; published patent application text for same-team prior disclosure and overlapping candidates..

Editorial assessment: Impact 2/5 · Validation 3/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

AI-ranked SARS-CoV-2 protein interactions guide discovery of an NSP10–NSP16 inhibitor

A machine-learning classifier applied to quantitative PPI assay data and AlphaFold-Multimer predictions to prioritize SARS-CoV-2 protein interactions; 15 of 34 high-confidence interactions were not previously listed in IMEx. The team then targeted the NSP10–NSP16 methyltransferase interface in an ultra-large virtual screen. Compound 459 bound NSP10 (MST affinity about 13 μM), reduced NSP10–NSP16 interaction in cells (BRET IC50 9.2 μM), inhibited complex methyltransferase activity by about 50% at 100 μM, and reduced SARS-CoV-2 reporter replication in cells (IC50 39.5 μM).

Technology: Multi-adaptive support-vector-machine (maSVM) classifier for quantitative protein–protein interaction data, coupled with AlphaFold-Multimer complex prediction; VirtualFlow/QuickVina2 and later docking methods performed compound screening

Family AlphaFold: profile · models page

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: A machine-learning classifier applied to quantitative PPI assay data and AlphaFold-Multimer predictions to prioritize SARS-CoV-2 protein interactions; 15 of 34 high-confidence interactions were not previously listed in IMEx.
  • Still unknown: The paper reports an experimental small-molecule hit, not a drug or clinical effect. Compound 459 has micromolar affinity and cellular antiviral IC50; authors explicitly state it requires extensive optimization. Methyltransferase inhibition was about 50% at 100 μM in one initial assay; subsequent data include cellular interaction and reporter-replication experiments. The NSP10–NSP16 interface was previously known and had prior virtual screening, but this paper reports a different compound scaffold and tested outcomes; shared target does not make this a duplicate. In vitro and cell assays do not demonstrate efficacy in animals or people.
  • AI role: The maSVM algorithm prioritized PPI interactions using quantitative LuTHy binary-assay data and AlphaFold-Multimer predictions, supporting selection and mapping of SARS-CoV-2 interactions, including the NSP10–NSP16 interface and interface hot spots. The drug-like compound hit itself was identified by VirtualFlow and docking against the selected interface, not generated by the classifier. Compound 459 was then experimentally tested for protein binding, interaction disruption, methyltransferase activity and viral replication.

Evidence sources: https://link.springer.com/article/10.1038/s44320-024-00019-8

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: in the top 10% of works of the same type, year and subfield (OpenAlex work metrics, 37 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Sequential model-based optimization discovers two previously unreported chemical reactions

A programmable robotic synthesis platform used inline HPLC, Raman or NMR feedback and closed-loop optimization to explore selected reaction spaces. In a tosMIC campaign, sequential model-based optimization with a Gaussian-process surrogate identified a reaction between toluenesulphonylmethyl isocyanide and benzylidenemalononitrile; subsequent optimization increased yield from the 27% initial condition to 49% (22 percentage points) over 50 reactions. The article also reports a second previously unreported reaction and an exploratory trifluoromethylation campaign that discovered new molecules.

Technology: ChemputationOptimizer, a sensor-integrated reaction engine with sequential model-based optimization (Gaussian-process surrogate); the platform also exposes random search, design-of-experiments, Bayesian-optimization and genetic-algorithm routines

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: The article also reports a second previously unreported reaction and an exploratory trifluoromethylation campaign that discovered new molecules.
  • Still unknown: The platform's reported yield gains are specific to these selected reaction spaces and objectives, not general chemistry performance. The tosMIC and phloroglucinol campaigns are reported as previously unreported reactions; product characterization and complete procedure details are in the article/supplement. The system includes several optimization algorithms and the authors note random search can be a strong baseline; the reported tosMIC optimization used SMBO with a Gaussian-process surrogate. No claim is made that the reactions are industrially useful or broadly generalizable. Primary source checked; no independent replication assessed.
  • AI role: For the tosMIC and phloroglucinol explorations, machine-guided combinatorial search proposed reaction parameters and inline spectroscopic feedback supplied product-yield measurements; a Gaussian-process sequential-model-based optimizer guided the reaction-space exploration and follow-up optimization. Human chemists defined reaction space and experimental constraints, and performed/characterized the experiments.

Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC10858227/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: in the top 10% of works of the same type, year and subfield (OpenAlex work metrics, 61 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Machine learning prioritizes ruthenium complexes with increased MRSA antibacterial hit rates

Researchers trained machine-learning models on 288 modularly synthesized ruthenium arene Schiff-base complexes and used them to prioritize 54 new compounds for antibacterial-activity testing. The model-selected set had a 53.7% hit rate against methicillin-resistant Staphylococcus aureus, compared with 9.4% in the original library, an approximately 5.7-fold enrichment.

Technology: Machine-learning activity-prediction models trained on ruthenium arene Schiff-base complex structures and antibacterial assay data

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Researchers trained machine-learning models on 288 modularly synthesized ruthenium arene Schiff-base complexes and used them to prioritize 54 new compounds for antibacterial-activity testing.
  • Still unknown: This is a study-specific in-vitro antibacterial hit-rate enrichment, not animal efficacy, clinical validation, or an approved antibiotic. The comparison is 54 model-selected compounds vs the 288-complex source library, not a randomized prospective comparison with all possible metal complexes.
  • AI role: Models were trained using assay data from 288 modularly synthesized ruthenium complexes and used to predict activity for a new set of 54 compounds. Those prioritized compounds were then assessed for antibacterial activity; the paper reports higher MRSA hit-rate in the model-selected set than in the original library.

Evidence sources: https://pubmed.ncbi.nlm.nih.gov/38088924/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: in the top 10% of works of the same type, year and subfield (OpenAlex work metrics, 50 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 2/5 · Validation 3/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Active learning selected oxidative methane coupling catalysts

The authors combined automatic feature engineering with active learning and high-throughput experiments for oxidative coupling of methane. Of 36 selected catalysts, 30 reached combined ethane and ethylene yields of at least 15% in the reported tests. This is a small-data, reaction-specific catalyst screening outcome; the paper also tests feature engineering on other catalysis datasets. A ChemRxiv version predates the journal publication, and this record does not claim journal-version priority.

Technology: Automatic feature engineering with active learning

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: The article reports 36 selected oxidative coupling of methane catalysts experimentally assessed; 30 achieved combined ethane and ethylene yields of at least 15% under the reported test conditions.
  • Still unknown: The paper also appeared as an earlier ChemRxiv version; this check did not establish whether every quantitative detail appeared in that version. The result is limited to the reported catalyst set and reaction conditions.
  • AI role: Automatic feature engineering extracted catalyst descriptors, and active learning selected catalysts for high-throughput experimental testing.

Evidence sources: https://d-nb.info/1324333405/34 · https://www.nature.com/articles/s42004-023-01086-y

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review date: 17 Nov 2026. Review scope: Full published article text in German National Library repository and publisher-indexed primary metadata/abstract; earlier ChemRxiv version noted but not compared..

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Active learning identifies experimentally improved ammonia catalysts

Literature-trained models and iterative experimental feedback guided a predetermined ammonia-catalyst search. Selected formulations exceeded many reported comparator activities in the authors' laboratory measurements. Initial literature-only predictions were insufficient, making feedback essential. This is a bounded formulation result, not demonstrated commercial green-ammonia performance or invention of promoted ruthenium catalysis.

Technology: Active-learning regression trained on literature and experimental feedback

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Researchers synthesized and tested model-selected catalyst formulations and updated the model.
  • Still unknown: Independent formulation priority, replication and industrial performance were not assessed.
  • AI role: Active-learning regression trained on literature and experimental feedback supported the stated study; experimental and interpretive work remained with researchers.

Evidence sources: https://pubs.rsc.org/en/content/articlelanding/2024/ta/d3ta05939a

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review date: 16 Nov 2026. Review scope: Indexed primary publisher abstract, author affiliations and publication metadata; prior frozen method review retained, author manuscript freshly blocked..

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Universidad ECCI

Machine learning screens cinnamic-acid derivatives for Leishmania enzyme inhibition

Two random-forest classifiers and docking ranked an in-house set of cinnamic-acid derivatives; selected compounds were tested against recombinant L. major DHFR-TS. The paper reports IC50 values of 7.5, 6.1 and 10.1 µM for three named natural products versus 1.4 µM for methotrexate. These are enzyme assays, not whole-parasite treatment results.

Technology: Random-forest classifiers; structure-based docking

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Primary article reports recombinant DHFR-TS inhibition for selected compounds and comparator.
  • Still unknown: Methods/results disagree on number of compounds; name/formula spellings and exact prior art need checking.
  • AI role: RF models and docking prioritize a small compound set for biochemical enzyme assays.

Evidence sources: https://doi.org/10.3390/molecules29010179

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review date: 16 Nov 2026. Review scope: PMC-hosted primary full text; publisher primary article page. Material checked for the specific claims stated in the record..

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Machine-learning-guided closed-loop synthesis experimentally realizes unreported dye molecules

A closed-loop molecular discovery platform combined machine-learning property prediction, automated synthesis and characterization, and iterative model updates to explore dye-like molecules. In the published Science abstract, the team reports experimentally realizing 294 previously unreported molecules across three design–make–test–analyze cycles on four rarely reported scaffolds, while also identifying six top-performing molecules in a second case study using pre-existing chemical-space examples.

Technology: Autonomous molecular-discovery platform integrating generative and property-prediction machine-learning models, Bayesian optimization, and synthesis-planning software with automated chemical synthesis and characterization

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: In the published Science abstract, the team reports experimentally realizing 294 previously unreported molecules across three design–make–test–analyze cycles on four rarely reported scaffolds, while also identifying six top-performing molecules in a second case study using pre-existing chemical-space examples.
  • Still unknown: This is molecular discovery in dye-like chemical systems, not a validated drug or biological effect. The 294 number is from the published Science abstract; an earlier ChemRxiv preprint describes 312 unreported molecules, so version-specific counts differ and the published figure is used here. The measurements addressed absorption wavelength, lipophilicity and photo-oxidative stability; no environmental, toxicological or commercial benefit is claimed. Primary-source abstract and author-lab description checked; independent replication not assessed.
  • AI role: Generative ML proposed candidate dye-like structures; property-prediction models ranked molecules against absorption, lipophilicity, and photo-oxidative-stability targets; automated multi-step synthesis and measurements generated training data for model retraining and the next cycles. The closed-loop process converted model-guided suggestions into synthesized and characterized compounds.

Evidence sources: https://doi.org/10.1126/science.adi1407

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: in the top 1% of works of the same type, year and subfield (OpenAlex work metrics, 195 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Coscientist: autonomous chemical research with language models

A GPT-4-guided system designs, plans and runs experiments using web search, documentation, code and laboratory automation, performing six tasks including optimization of palladium cross-coupling reactions. The abstract does not make clear which tasks used real hardware; only the abstract was reviewed.

Technology: Coscientist (GPT-4)

Family GPT-4: profile · models page

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: The system was run on six tasks, including a palladium-catalysed cross-coupling reaction optimization that the abstract reports as successful.
  • Still unknown: The abstract does not state which tasks used physical hardware and which used simulation, and states no limitations.
  • AI role: GPT-4 planned and executed experiments by using web search, documentation, code execution and experimental automation tools.

Evidence sources: https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=DOI:10.1038/s41586-023-06792-0&resultType=core&format=json

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: in the top 1% of works of the same type, year and subfield (OpenAlex work metrics, 1811 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 4/5 · Validation 3/5 · Novelty 5/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Extrapolative machine learning identifies a multielement RWGS catalyst

In a closed loop combining machine-learning prediction with experiments, researchers tested 300 reverse water-gas shift catalyst formulations over 44 cycles, starting from 45 data points. They report Pt(3)/Rb(1)-Ba(1)-Mo(0.6)-Nb(0.2)/TiO₂ as the optimal tested composition; Nb was absent from the initial dataset. The result is specific to the study’s catalyst tests and conditions, not commercial CO₂ conversion. The earlier ChemRxiv version is not used to claim priority here.

Technology: Extrapolative machine learning; closed-loop prediction and experiment

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: The source reports 300 experimentally tested catalysts over 44 ML-plus-experiment cycles and identifies the best tested formulation.
  • Still unknown: Commercial operation, independent replication and performance beyond reported experimental conditions are not established.
  • AI role: Machine-learning predictions guide sequential catalyst selection; experimental results update the closed-loop search.

Evidence sources: https://www.nature.com/articles/s41467-023-41341-3.pdf

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review date: 16 Nov 2026. Review scope: Publisher PDF indexed primary full text and publication metadata; direct HTML endpoint blocked by cookie redirect..

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 4/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

PhotoCat guides experimental discovery of photocatalytic reactions

The PhotoCat Transformer was pretrained on general reaction data and fine-tuned on the authors’ photocatalysis database to predict products and reaction conditions. In the 2023 ChemRxiv version, the authors report using it to guide experimental validation of previously unreported photocatalytic reactions. The later 2026 journal version describes four novel validated reactions, whereas the preprint reports five; this entry makes no exact-count claim and treats the journal article as a revised version of the same study.

Technology: Transformer-based deep-learning reaction predictor pretrained on USPTO and fine-tuned on PhotoCatDB, with reaction-condition inputs

Publication: Preprint · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Preprint
  • Test type: direct test
  • Tested: The PhotoCat Transformer was pretrained on general reaction data and fine-tuned on a photocatalysis database to predict products and conditions. The 2023 ChemRxiv version reports wet-lab validation of five predicted reactions; the 2026 journal abstract reports four novel validated reactions.
  • Still unknown: The preprint and journal version differ in their reported count of validated novel reactions. This record states the shared model-guided experimental discovery claim and makes no exact-count claim.
  • AI role: A Transformer reaction predictor was pretrained on USPTO and fine-tuned on PhotoCatDB to rank photocatalytic reaction outcomes and conditions, guiding experimental validation.

Evidence sources: https://chemrxiv.org/engage/api-gateway/chemrxiv/assets/orp/resource/item/65004fd999918fe537d200ab/original/photo-cat-an-artificial-intelligence-driven-synthesis-planning-platform-for-photocatalysis.pdf · https://pmc.ncbi.nlm.nih.gov/articles/PMC12913628/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review date: 17 Nov 2026. Review scope: Primary ChemRxiv v2 full-text PDF and version-history metadata; primary PMC full-text journal version and publication metadata. The preprint and journal versions report different counts (five versus four); the summary preserves this discrepancy..

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Machine learning guides Fe–Mn–Ni catalysts for NOx reduction

An artificial neural network trained on literature data and a genetic algorithm helped identify Fe–Mn–Ni catalyst compositions for ammonia selective catalytic reduction. Researchers synthesized and tested model-guided candidates and report good NOx conversion for Fe₀.₃₁₉Mn₀.₂₀₆Ni₀.₄₇₅ over the tested temperature range. The work combines model selection, human screening and controlled experiments; the ratio is not a complete oxide formula and does not demonstrate industrial exhaust performance. The article text differs on whether the workflow involved two or four iterations, so this summary avoids an iteration count.

Technology: Artificial neural network and genetic algorithm

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: The paper reports synthesis and catalytic conversion testing of selected Fe–Mn–Ni formulations.
  • Still unknown: Article reports inconsistent iteration counts; test range and lab results do not establish durability or practical flue-gas performance.
  • AI role: An ANN and genetic search ranked candidate Fe–Mn–Ni compositions; experimental results were used in the iterative optimization workflow.

Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC10666265/ · https://pubmed.ncbi.nlm.nih.gov/37393584/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review date: 16 Nov 2026. Review scope: Primary PMC full text, PubMed metadata/affiliation and publisher date chronology..

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Sequence-conditioned CogMol designs SARS-CoV-2 spike and main-protease inhibitors

IBM Research and academic collaborators used the CogMol deep generative foundation model to design compounds from target protein sequence information alone for SARS-CoV-2 spike receptor-binding domain and main protease. In vitro inhibition was reported for two of four synthesized candidates for each target; the most potent spike-RBD inhibitor also showed activity against multiple variants in live-virus neutralization assays.

Technology: CogMol, a sequence-conditioned deep generative foundation model using a molecular variational autoencoder and protein–ligand binding predictor

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: In vitro inhibition was reported for two of four synthesized candidates for each target; the most potent spike-RBD inhibitor also showed activity against multiple variants in live-virus neutralization assays.
  • Still unknown: The source describes a proof of concept with a small selected synthesis set from roughly 100 ranked designs. Only in-vitro target assays were performed; the leading RBD inhibitor was also tested in live-virus neutralization against variants. The paper says ADME attributes were not experimentally validated and further work is needed to confirm inhibitor binding mechanisms. These are preclinical leads, not medicines or demonstrated clinical benefit.
  • AI role: The team sampled CogMol conditioned on each target's amino-acid sequence to generate candidate small molecules; candidates were filtered and ranked in silico, synthesized and assayed. The model generated compounds without target-specific fine-tuning or target-structure input at inference; assays found micromolar inhibition for 2/4 synthesized candidates for each target.

Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC10284550/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: in the top 10% of works of the same type, year and subfield (OpenAlex work metrics, 66 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Machine learning guides direct synthesis of higher-silica faujasite

Machine-learning models learned relationships between crystallization composition/conditions and zeolite characteristics from prior synthesis records, guiding human selection of conditions for faujasite synthesis. The team reports directly synthesized, unseeded, organic-structure-directing-agent-free FAU with Si/Al 3.5, compared with the prior reported maximum 2.8. Following ion exchange, laboratory tests report improved propane cracking and dehydrogenation rates per proton site relative to the lower-silica comparator. The FAU framework was already known; this result concerns composition and route, not a new framework or industrial performance.

Technology: Geometric harmonics, neural networks and Gaussian-process regression

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Researchers synthesized and characterized FAU and tested acid catalysts in propane cracking/dehydrogenation under laboratory conditions.
  • Still unknown: Known framework; activity comparison is assay-specific and does not establish industrial gains.
  • AI role: Models trained on prior synthesis records identified composition/condition settings that guided direct high-silica FAU synthesis.

Evidence sources: https://authors.library.caltech.edu/records/makq9-r3046

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review date: 16 Nov 2026. Review scope: Primary Caltech institutional record and deposited published article/supplement; date, affiliations, methods and abstract/results..

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Machine-learning screening yields novel potent acetylcholinesterase and butyrylcholinesterase inhibitors

NCATS researchers trained machine-learning QSAR models and applied them to a roughly 360,000-compound library. Experimental colorimetric and orthogonal fluorescence assays confirmed new AChE/BChE inhibitors; the study reports 22 AChE and 67 BChE inhibitors with IC50 below 5 μM, including novel structural series, with some submicromolar compounds.

Technology: Ensemble comparison of machine-learning QSAR classifiers (including XGBoost, random forest, SVM, neural network and naïve Bayes)

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Experimental colorimetric and orthogonal fluorescence assays confirmed new AChE/BChE inhibitors; the study reports 22 AChE and 67 BChE inhibitors with IC50 below 5 μM, including novel structural series, with some submicromolar compounds.
  • Still unknown: These are biochemical inhibitor assays and early medicinal-chemistry leads; they do not establish brain exposure, in-vivo efficacy, safety or clinical benefit. Several known compounds were newly identified as inhibitors, so the count of all assay hits should not be described as entirely novel molecules.
  • AI role: Models trained on quantitative high-throughput screening data prioritized compounds in a ~360,000-compound collection for AChE/BChE inhibition and selectivity assays. Researchers tested 2,762 high-scoring compounds in colorimetric assays and used an orthogonal fluorescence assay to confirm predicted inhibitors; subsequent structure–activity analysis identified novel scaffolds.

Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC10688023/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: in the top 10% of works of the same type, year and subfield (OpenAlex work metrics, 26 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

AI-generated catalyst design led to experimentally tested polymerization catalysts

A regression-transformer model fine-tuned on historical ring-opening-polymerization data generated candidate catalysts. Researchers selected synthetically accessible, modified surrogates of one model-generated catalyst for laboratory testing. One surrogate, 5c, reached 60% conversion in 32 minutes for monomer 2a and 100% conversion in 45 minutes for monomer 2d, with reported dispersities of 1.06 and 1.34, respectively. The first ChemRxiv v1 (2022-06-27) described computational generation and recovery of a literature-known catalyst. The experimental results appeared in ChemRxiv v2, posted 2023-02-13, and the 2023 peer-reviewed journal article followed on 2023-06-21. The measured tests concern modified surrogates, not the unmodified generated structure.

Technology: Regression Transformer generative model fine-tuned on historical polymerization data represented with Chemical Markdown Language

Publication: Preprint · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Preprint
  • Test type: direct test
  • Tested: The journal reports model-derived catalyst 3g and synthetically accessible modified surrogates 5c/5d. Surrogate 5c produced 60% conversion of monomer 2a in 32 minutes (dispersity 1.06) and 100% conversion of monomer 2d in 45 minutes (dispersity 1.34).
  • Still unknown: The tested catalysts were modified surrogates rather than the unmodified model-generated 3g structure. These are laboratory ring-opening-polymerization outcomes, not process-scale or commercial performance. The earlier 2022 v1 reported model generation and literature-known catalyst recovery; the experimentally validated result is reported in v2 on 2023-02-13 and in the later journal article.
  • AI role: A regression transformer generated catalyst 3g from historical polymerization data; researchers tested synthetically accessible modified surrogates 5c and 5d that preserved its endocyclic guanidine motif.

Evidence sources: https://chemrxiv.org/engage/chemrxiv/article-details/62b60865e84dd185e60214af · https://chemrxiv.org/engage/chemrxiv/article-details/63e668409da0bc6b33b1870c · https://pmc.ncbi.nlm.nih.gov/articles/PMC10284867/ · https://pubmed.ncbi.nlm.nih.gov/37344485/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review date: 17 Nov 2026. Review scope: Primary ChemRxiv v1/v2 records and linked full v1 PDF for version chronology; primary PubMed Central journal full text for model-to-catalyst relationship and measured outcomes; PubMed metadata for journal publication date and affiliations..

Editorial assessment: Impact 2/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Support-vector regression selects more selective tetrahydropyran catalysts

Models of catalyst fragments and reaction settings selected catalysts for asymmetric tetrahydropyran synthesis. A tested catalyst reached an enantiomeric ratio of 91.5:8.5, above the training examples' maximum of 82:18. The result improves a known catalyst class for a focused reaction; it does not establish invention of the reaction or broad generality.

Technology: Support-vector regression with tunable fragment descriptors

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Previously untested catalyst substituents were prepared and experimentally evaluated.
  • Still unknown: Exhaustive catalyst prior art, independent replication and broader reaction scope were not assessed.
  • AI role: Support-vector regression with tunable fragment descriptors supported the stated study; experimental and interpretive work remained with researchers.

Evidence sources: https://onlinelibrary.wiley.com/doi/full/10.1002/anie.202218659

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review date: 16 Nov 2026. Review scope: Primary publisher article dates, affiliations and selected model/experimental-validation Results; supplementary data not reviewed..

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Machine-learning-guided electrochemical screening isolates five new reaction products

A machine-learning-guided, chemist-in-the-loop screen predicted electrochemical reactivity for 38,865 potential partners with 1,4-dicyanobenzene (DCB). Automated tests of 20 prioritized, previously untested partners yielded DCB conversion in 16; five produced an isolable major product judged of synthetic or mechanistic interest. The reported examples include a silyl-enol-ether coupling and electrochemical benzylic arylation of benzyltrimethylsilane with DCB, plus three mechanistically unexpected products.

Technology: Machine-learning-guided reaction-competency classifier using a DFT-informed molecular representation

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Automated tests of 20 prioritized, previously untested partners yielded DCB conversion in 16; five produced an isolable major product judged of synthetic or mechanistic interest.
  • Still unknown: This is an early reaction-discovery demonstration under one fixed DCB electrolysis system, not a general reaction predictor or optimized synthetic platform. Seven of the 16 DCB-converting selected reactions gave complex mixtures without an appreciable single isolated product; five yielded isolable major products, reported at low yields and without method optimization. DCB conversion alone does not establish a product identity, and the three mechanistically ambiguous outcomes require further study.
  • AI role: The model screened a 38,865-member virtual set for reaction competency and, with chemist selection, prioritized 20 new partners for automated electrolysis. The experiments measured DCB conversion, then product analysis isolated and characterized five major products from the selected set.

Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC9756344/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: in the top 10% of works of the same type, year and subfield (OpenAlex work metrics, 78 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Prospective machine-learning screen identifies six previously unreported OATP1B1 inhibitors

Researchers trained and compared seven machine-learning classifiers for OATP1B1 transporter inhibition. In prospective CHO-cell testing, 15 of 19 compounds predicted active by SVC models inhibited the transporter; six (abamectin, asiaticoside, berbamine, doramectin, mobocertinib and umbralisib) were described as previously unreported OATP1B1 inhibitors.

Technology: Seven molecular-fingerprint machine-learning classifiers, with prospective selection by support vector classifier (SVC)

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: In prospective CHO-cell testing, 15 of 19 compounds predicted active by SVC models inhibited the transporter; six (abamectin, asiaticoside, berbamine, doramectin, mobocertinib and umbralisib) were described as previously unreported OATP1B1 inhibitors. The new candidate screen used contracted Eurofins assays; external execution is not a separate replication study.
  • Still unknown: The 84% refers to a small prospective test set of 19 compounds, and transporter inhibition was measured in vitro in engineered CHO cells. These findings flag possible drug–drug interaction liability; they do not show clinical interaction or patient outcome. The authors qualify six inhibitors as appearing novel at publication.
  • AI role: Models trained on in-vitro OATP1B1 inhibition data ranked compounds for prospective testing. Researchers tested 19 model-selected compounds in transporter-expressing CHO cells and observed inhibition for 15, including six compounds the authors described as novel OATP1B1 inhibitors.

Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC9873312/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: at percentile 89 of works of the same type, year and subfield (OpenAlex work metrics, 31 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 2/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Generative learning selects experimentally active EGFR inhibitor analogs

A recurrent generative model refined by reinforcement learning selected compounds for an EGFR enzyme assay. Four of 15 predicted-active compounds were active, including two with nanomolar inhibition; five predicted-inactive controls were inactive. The experimentally tested compounds were commercially obtainable analogs of a known privileged scaffold, not a wholly new drug scaffold or demonstrated cellular therapy.

Technology: Recurrent neural generator with reinforcement learning

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: An independent contract assay tested selected compounds and negative controls.
  • Still unknown: Cellular efficacy, selectivity, safety and novel-scaffold discovery are not established.
  • AI role: Recurrent neural generator with reinforcement learning supported the stated study; experimental and interpretive work remained with researchers.

Evidence sources: https://www.nature.com/articles/s42004-022-00733-0 · https://www.nature.com/articles/s42004-022-00733-0.pdf

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review date: 16 Nov 2026. Review scope: Primary publisher PDF abstract, selected experimental Results and affiliations; primary Crossref online-date metadata. Supplement not reviewed..

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

LIGHTHOUSE identifies riboflavin 5′-monophosphate as a PPAT inhibitor

LIGHTHOUSE, a deep-learning model using protein sequence and chemical representations, nominated riboflavin 5′-monophosphate as an inhibitor of PPAT, a target without reported inhibitors in the paper. A biochemical assay measured PPAT inhibition with an IC50 of 7.5 μM, close to the model’s predicted range; the same study also tested predicted antiviral compounds in cell models.

Technology: LIGHTHOUSE (ensemble graph-based deep-learning platform)

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: LIGHTHOUSE, a deep-learning model using protein sequence and chemical representations, nominated riboflavin 5′-monophosphate as an inhibitor of PPAT, a target without reported inhibitors in the paper.
  • Still unknown: The result is an in-vitro enzyme assay, not a validated anticancer treatment. The authors note LIGHTHOUSE can produce false positives; COVID-19 results were also cell-based and had no clinical evidence.
  • AI role: The model scored protein–compound relationships without requiring a 3D target structure and prioritized the compound from >333,000 candidates for experimental follow-up. The authors assayed PPAT activity and confirmed inhibition in vitro.

Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC9549714/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: at percentile 57 of works of the same type, year and subfield (OpenAlex work metrics, 9 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 2/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Deep-learning screening identifies covalent SARS-CoV-2 main-protease inhibitor scaffolds

A two-stage deep-learning screen selected cysteine-directed compounds for testing against SARS-CoV-2 3CL protease. Of 32 top-ranked compounds experimentally assayed, 12 inhibited the enzyme; six had IC₅₀ values below 12 μM, with the strongest reported at 1.4 μM. Five of those six exhibited time-dependent inhibition consistent with covalent binding.

Technology: Deep-learning COVCL and NOVCL classifiers integrating a directed message-passing neural network and feed-forward network

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: A two-stage deep-learning screen selected cysteine-directed compounds for testing against SARS-CoV-2 3CL protease.
  • Still unknown: Evidence in the primary abstract is biochemical enzyme inhibition and inhibition-mechanism assays. It does not establish antiviral efficacy in cells/animals or clinical benefit. Potency values are for the reported enzyme assay; most tested compounds were inactive, and the six sub-12 μM hits are early leads.
  • AI role: The COVCL model first prioritized cysteine-targeted covalent candidates, then NOVCL ranked those candidates for predicted 3CL-protease inhibition. Researchers tested 32 selected molecules in enzyme assays and performed follow-up tests of time-dependent inhibition for active hits.

Evidence sources: https://pubmed.ncbi.nlm.nih.gov/36209629/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: at percentile 86 of works of the same type, year and subfield (OpenAlex work metrics, 25 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 2/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Machine learning generates active daunomycin-binding synthetic nucleic-acid polymers outside selected sequence families

The study combined in-vitro selection with a conditional variational autoencoder trained on selection data to generate high-affinity highly functionalized nucleic-acid polymer (HFNAP) sequences. Experimentally generated CVAE sequences bound daunomycin with reported dissociation constants of 9–26 nM despite having no apparent primary-sequence similarity to experimentally selected polymers; competition selections found selected generated variants performed as well as or better than sequences obtained after multiple rounds of high-stringency selection.

Technology: Conditional variational autoencoder (CVAE) trained on in-vitro selection-derived HFNAP sequence/fitness data

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Experimentally generated CVAE sequences bound daunomycin with reported dissociation constants of 9–26 nM despite having no apparent primary-sequence similarity to experimentally selected polymers; competition selections found selected generated variants performed as well as or better than sequences obtained after multiple rounds of high-stringency selection.
  • Still unknown: The finding is target binding by synthetic HFNAPs, not therapeutic efficacy or cellular activity. Binding results are specific to daunomycin and reported assays; the authors describe sequences as having no apparent similarity rather than proving wholly unrelated functional mechanisms. Some competition performance was assessed by in-vitro selection rather than a clinical or organism-level test.
  • AI role: The authors used the CVAE to condition generation of novel HFNAP sequences on predicted daunomycin-binding fitness, then synthesized/tested generated sequences in binding and competition-selection assays. The machine-learning model generated the active sequences; wet-lab selection and competition assays established their binding activity.

Evidence sources: https://www.nature.com/articles/s41467-022-31955-4

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: at percentile 79 of works of the same type, year and subfield (OpenAlex work metrics, 23 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Learned subgroup rules guide higher-component oxidation catalysts

Rules learned from high-throughput catalyst measurements guided tests of more complex four- and five-component materials for propylene oxidation. Selected materials reached approximately 60% oxygenate yield in laboratory tests. The learned descriptor ranges are associative design guides, not proven atomic mechanisms or universal industrial performance rules.

Technology: Subgroup-discovery rule learning

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: More complex catalyst compositions were tested against rules learned from three-component measurements.
  • Still unknown: Independent replication and industrial durability were not assessed; temperature rules did not transfer unchanged.
  • AI role: Subgroup-discovery rule learning supported the stated study; experimental and interpretive work remained with researchers.

Evidence sources: https://doi.org/10.1021/acscatal.1c04793 · https://pure.mpg.de/pubman/item/item_3369647_4/component/file_3370185/acscatal.1c04793.pdf

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review date: 16 Nov 2026. Review scope: Indexed primary publisher text and deposited publisher PDF abstract and selected experimental-validation Results; supplements not reviewed..

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Machine learning identifies new CYP2C9-inhibiting drugs and drug metabolites

A structure-, dynamics- and physicochemical-feature machine-learning approach predicted CYP2C9 inhibition and experimentally identified four drugs—vatalanib, piriqualone, ticagrelor and cloperidone—as strong inhibitors, plus sertindole, asapiprant, duvelisib and dasatinib as moderate inhibitors in reported assays. Vatalanib had a reported CYP2C9 inhibition IC50 of 0.067 μM. Separate metabolism assays characterized CYP2C9-produced metabolites for abemaciclib, cloperidone, vatalanib and tarafenacin.

Technology: Machine-learning inhibitor predictors combining CYP2C9 structural/dynamics information with inhibitor physicochemical properties

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: A structure-, dynamics- and physicochemical-feature machine-learning approach predicted CYP2C9 inhibition and experimentally identified four drugs—vatalanib, piriqualone, ticagrelor and cloperidone—as strong inhibitors, plus sertindole, asapiprant, duvelisib and dasatinib as moderate inhibitors in reported assays.
  • Still unknown: Inhibition and metabolite findings are from the reported in vitro systems and do not by themselves establish clinically relevant drug–drug interactions, exposure changes, or patient harm. Reported inhibition strengths are assay-specific; clinical relevance depends on dose, exposure and context and requires pharmacokinetic/clinical assessment.
  • AI role: The authors used machine-learning models to prioritize drug-like compounds for CYP2C9 inhibition testing. Follow-up in vitro assays identified the listed strong and moderate inhibitors; CYP2C9 metabolism assays separately characterized metabolites of four drugs.

Evidence sources: https://pubmed.ncbi.nlm.nih.gov/35081108/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: in the top 10% of works of the same type, year and subfield (OpenAlex work metrics, 38 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 2/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Retrosynthetic software finds experimentally validated supply routes for umifenovir

Retrosynthesis software proposed alternative chemical supply routes for the antiviral umifenovir and eleven other antiviral and anti-inflammatory drugs. The authors experimentally validated four routes to umifenovir and one route to bromhexine, one of which converts six C-H bonds. The work is a proof of route feasibility for supply-chain resilience, not a new therapeutic claim.

Technology: Computer-aided retrosynthesis software with exclusion of known starting materials

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: The authors experimentally validated four routes to umifenovir and one route to bromhexine proposed by the software.
  • Still unknown: Not reported.
  • AI role: Retrosynthesis software proposed alternative supply routes for umifenovir and other antiviral and anti-inflammatory drugs.

Evidence sources: https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=DOI:10.1038/s41467-021-27547-3&resultType=core&format=json

The legacy label “Not reported” does not distinguish information absent from reviewed material from information not assessed. It is not a scientific finding of absence.

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: at percentile 87 of works of the same type, year and subfield (OpenAlex work metrics, 38 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Google DeepMind

DM21: a neural network as a density-functional theory functional

A neural network represents the functional used in DFT and addresses two longstanding errors of traditional functionals: excessive electron delocalization and spin-symmetry breaking during bond breaking. Published in Science with open code; the post gives no benchmark figures and DFT remains an approximation.

Technology: DM21 (neural network for the exchange-correlation functional)

Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: retrospective or benchmark
  • Tested: DM21 is compared with traditional functionals on benchmarks for main-group atoms and molecules and on bond-breaking and charge-localisation examples.
  • Still unknown: The page gives no numerical benchmark figures and says DFT remains an approximation.
  • AI role: Neural network represents the exchange-correlation functional in density functional theory.

Evidence sources: https://deepmind.google/discover/blog/simulating-matter-on-the-quantum-scale-with-ai/ · https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=DOI%3A10.1126/science.abj6511&resultType=core&format=json

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: blog. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: in the top 1% of works of the same type, year and subfield (OpenAlex work metrics, 416 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 3/5 · Validation 2/5 · Novelty 4/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Unsupervised machine learning identifies palladium catalysts that form dinuclear complexes

A workflow built on only five experimental data points screened 348 phosphine ligands and predicted ligands that favour dinuclear Pd(I) complexes over Pd(0) and Pd(II) species. The authors experimentally verified the predictions, including ligands that had never been synthesized before. The evidence is limited to the Pd speciation system studied, not a general catalyst performance gain.

Technology: Unsupervised machine learning workflow with clustering on generalized parameter databases and in silico data

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: The authors experimentally verified the predicted phosphine ligands, including ligands never synthesized before, and they gave dinuclear Pd(I) complexes. The ligand predictions were checked by laboratory experiments.
  • Still unknown: Not reported.
  • AI role: Unsupervised machine learning workflow built on five experimental data points, screened 348 phosphine ligands and predicted those favouring dinuclear Pd(I) complexes.

Evidence sources: https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=DOI:10.1126/science.abj0999&resultType=core&format=json

The legacy label “Not reported” does not distinguish information absent from reviewed material from information not assessed. It is not a scientific finding of absence.

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: in the top 10% of works of the same type, year and subfield (OpenAlex work metrics, 167 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 4/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Neural-network-guided autonomous experiments reveal an unreported TosMIC reaction

A reactivity-seeking neural network coupled to an autonomous reaction platform navigated unknown chemical mixtures using real-time analytical data. Among the reactions it identified, follow-up characterization revealed an unreported photochemical reaction and a new reactivity mode for p-toluenesulfonylmethyl isocyanide (TosMIC): six equivalents of TosMIC underwent a multistep single-substrate cascade yielding a trimeric product in 47% unoptimized yield and forming five C–C bonds.

Technology: Reactivity-seeking neural network and junction-tree neural network integrated with an autonomous robotic synthesis and analytics platform

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Among the reactions it identified, follow-up characterization revealed an unreported photochemical reaction and a new reactivity mode for p-toluenesulfonylmethyl isocyanide (TosMIC): six equivalents of TosMIC underwent a multistep single-substrate cascade yielding a trimeric product in 47% unoptimized yield and forming five C–C bonds.
  • Still unknown: The source describes several found reactions, some previously known; this entry is limited to the distinct unreported TosMIC reactivity result. The reported 47% yield was unoptimized. The platform identified candidate reactions, while follow-up analyses and manual chemical investigation established the product and mechanism; no broad claim of fully autonomous mechanistic proof is made.
  • AI role: A convolutional neural network classified reaction signatures from online NMR/IR/analytical data and a junction-tree neural network encoded reagent structures to help prioritize new experiments. The closed-loop system surfaced reactions; chemists then manually investigated and characterized selected outcomes, including the TosMIC trimer-forming cascade.

Evidence sources: https://pubmed.ncbi.nlm.nih.gov/34849401/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: in the top 10% of works of the same type, year and subfield (OpenAlex work metrics, 82 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Machine-intelligence design yields selective synthetic COX-1 inhibitors from a marine natural-product template

Using Marinopyrrole A as a template, an automated design workflow generated and synthesized new natural-product-inspired molecules; target-prediction software pointed to cyclooxygenases. In biochemical assays, synthetic mimetics 2 and 2a inhibited COX-1 at 0.10 and 0.16 μM, respectively, and compound 2a was crystallographically observed bound to COX-1. Human platelet assays also showed inhibition.

Technology: DOGS rule-based de novo molecule construction and CATS molecular-similarity ranking, coupled with SPiDER self-organizing-map machine-learning target prediction

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: In biochemical assays, synthetic mimetics 2 and 2a inhibited COX-1 at 0.10 and 0.16 μM, respectively, and compound 2a was crystallographically observed bound to COX-1.
  • Still unknown: The study tests a small number of designed molecules; automated structure generation was rule-based and similarity-driven rather than a trained generative neural model. The COX-1 effects were biochemical/platelet assays, not clinical effects; the platelet IC50s for 2, 2a, 2b were 9, 13 and 56 nM, but COX-2 inhibition occurred at higher concentrations and selectivity varies by assay. Do not call these approved drugs or imply therapeutic benefit.
  • AI role: DOGS algorithmically generated natural-product-inspired structures and suggested synthetic routes; CATS ranked candidates by pharmacophore similarity. SPiDER's cascaded self-organizing-map models predicted molecular target families, including COX. Experimental target assays confirmed the previously unrecognized COX-1 activity of selected synthetic compounds, with crystallographic confirmation for compound 2a.

Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC8373093/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: at percentile 84 of works of the same type, year and subfield (OpenAlex work metrics, 31 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

ETH Zurich

Generative AI and on-chip synthesis produce new LXR agonists

A recurrent neural network generated de novo LXR agonist designs, and 25 were synthesized in a microfluidic flow system. In vitro screening found 17 hits (68%), and 12 were confirmed as potent LXR agonists after batch resynthesis. Validation is in vitro only, with no in vivo testing, and the sources disagree on the number of designs retained for synthesis.

Technology: LSTM recurrent neural network generator with transfer learning, coupled to on-chip flow synthesis

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: 25 synthesized designs were screened in reporter gene assays (17 hits); 12 were confirmed as LXR agonists after batch resynthesis, with EC50 0.18 to 4.5 micromolar for LXR-alpha. No in vivo testing was reported.
  • Still unknown: Chemical space was limited to one-step reactions, the primary screen used crude mixtures, two compounds were not confirmed, and the authors describe the work as proof-of-concept.
  • AI role: A generative LSTM pretrained on 656,070 compounds and fine-tuned on 40 known LXR-alpha agonists, with a virtual reaction filter of 17 reaction schemes.

Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC8195470/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: in the top 1% of works of the same type, year and subfield (OpenAlex work metrics, 203 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 4/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Machine learning guided Cu catalyst additive discovery for CO₂ reduction

Across three iterative cycles of experimental testing, machine-learning analysis, prediction and redesign, the authors obtained Cu catalysts selective for CO, formic acid and C₂+ products in electrochemical CO₂ reduction. This summary follows the primary abstract and does not assign disputed sample identifiers or report cell-specific percentages.

Technology: Machine-learning-guided iterative additive screening and redesign

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: The primary abstract reports three iterative cycles of experimental testing and model-guided redesign, producing Cu catalysts selective for CO, formic acid and C₂+ products. No disputed sample-level metrics are included.
  • Still unknown: Full text and supporting information were not reviewed. The exact sample-to-performance mapping and the third-party rendering’s sample-number discrepancy remain unverified.
  • AI role: Machine-learning analysis of experimental additive/catalyst data guided predictions and redesign over three iterative experimental cycles.

Evidence sources: https://pubs.acs.org/doi/abs/10.1021/jacs.1c00339

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review date: 17 Nov 2026. Review scope: Primary ACS abstract, author-affiliation and article-history metadata; full text and supporting information not reviewed..

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Google DeepMind

FermiNet: quantum physics of molecules from first principles

FermiNet approximates molecular ground-state electron wavefunctions from first principles. Against the coupled-cluster reference, it recovered 99.8% of the correlation energy beyond a single Slater determinant for the smallest molecules and at least 97% for bicyclobutane (30 electrons). It also modeled stretched nitrogen and hydrogen chains. These are computational results from the original work announced on 19 October 2020. The source page was updated on 22 August 2024 with a separate excited-state study; that later study's carbon-dimer error is not a 2020 result.

Technology: FermiNet

Publication: Publication status unspecified in annotation · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Not reported
  • Test type: retrospective or benchmark
  • Tested: Energies compared with coupled-cluster and variational Monte Carlo reference calculations: 99.8% of the correlation energy beyond a single determinant recovered for the smallest molecules and at least 97% for bicyclobutane.
  • Still unknown: The coupled-cluster baseline scales poorly, so results apply to medium-sized molecules; the Monte Carlo method is approximate; excited-state results are within about 0.1 eV of the most demanding calculations rather than matching them.
  • AI role: Neural network wavefunction, antisymmetric in its inputs, trained by minimizing energy with Monte Carlo sampling using only nuclear positions.

Evidence sources: https://deepmind.google/blog/ferminet-quantum-physics-and-chemistry-from-first-principles/

The legacy label “Not reported” does not distinguish information absent from reviewed material from information not assessed. It is not a scientific finding of absence.

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: blog. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: in the top 1% of works of the same type, year and subfield (OpenAlex work metrics, 664 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 3/5 · Validation 2/5 · Novelty 4/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Chematica: computational synthesis planning for natural products

The system plans multistep syntheses of complex natural products, with routes in a Turing-test study largely indistinguishable from human routes; three computer-designed syntheses were carried out in the lab: dauricine, tacamonidine, and lamellodysidine A. Based on over 100,000 rules plus AI, it is not modern generative AI.

Technology: Chematica / Allchemy

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: In a Turing-like test, synthesis experts largely could not distinguish the program's routes from human ones; three computer-designed natural-product syntheses were carried out in the lab.
  • Still unknown: How often its routes fail across other targets is not reported in the abstract.
  • AI role: The planning program combines chemistry knowledge, data-based AI routines and causal relationships to plan several steps ahead.

Evidence sources: https://doi.org/10.1038/s41586-020-2855-y

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: in the top 10% of works of the same type, year and subfield (OpenAlex work metrics, 295 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 4/5 · Validation 3/5 · Novelty 4/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

A mobile robotic chemist

A mobile robot ran 688 experiments in eight days in a ten-variable space, guided by a batch Bayesian algorithm, to search for photocatalysts that produce hydrogen from water; the best mixtures were six times more active than the initial formulations. This optimizes a formulation, not a new structure; only the abstract was reviewed.

Technology: Batch Bayesian optimization and mobile robot

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: 688 autonomous experiments over eight days in a ten-variable space; the best photocatalyst mixtures were about six times more active than the initial formulations.
  • Still unknown: The abstract lists no limitations of this study; it notes that atomistic simulations cannot yet capture multi-scale complexity and that robots are hard to adopt widely across diverse samples and instruments.
  • AI role: A batch Bayesian algorithm chose which experiments the mobile robot ran next, automating the researcher rather than the instruments.

Evidence sources: https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=DOI:10.1038/s41586-020-2442-2&resultType=core&format=json

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: in the top 1% of works of the same type, year and subfield (OpenAlex work metrics, 1704 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 4/5 · Validation 3/5 · Novelty 5/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

DEL-trained models identify experimentally active small-molecule hits

Models trained on DNA-encoded-library selection data prioritized roughly 2,000 compounds for biochemical testing across three targets. The study reports active and potent hits outside the training libraries. Human filtering and chemical review contributed to selection. These are biochemical hits, not established safe or effective drugs; the authors disclose commercial interests and related patent applications.

Technology: Graph-convolutional neural networks and random forests

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Newly selected compounds were tested in biochemical assays.
  • Still unknown: Cellular efficacy, selectivity, safety and clinical benefit remain unestablished.
  • AI role: Graph-convolutional neural networks and random forests supported the stated study; experimental and interpretive work remained with researchers.

Evidence sources: https://accio.github.io/AMIDD/assets/2020/13-14/McCloskey-2020-ML-DELT.pdf

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review date: 16 Nov 2026. Review scope: Author-hosted published-article PDF abstract, selected experimental Results, affiliations and disclosures; prior PubMed online date retained..

Editorial assessment: Impact 3/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Cu–Al electrocatalysts for CO2 found with active learning

DFT and active machine learning identified Cu–Al catalysts that reduce CO2 to ethylene with over 80% Faradaic efficiency at 400 mA/cm² (about 66% for pure copper) and 55% half-cell energy efficiency at 150 mA/cm². Synthesized and tested in the lab; the abstract does not state how many candidates were screened.

Technology: Active machine learning and DFT

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

Result, AI contribution and limits

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

  • Publication status: Peer-reviewed paper
  • Test type: direct test
  • Tested: Cu-Al catalysts synthesized and tested in the lab: over 80% Faradaic efficiency for ethylene at 400 mA/cm2 versus about 66% for copper, and 55% half-cell energy efficiency at 150 mA/cm2; in situ X-ray absorption examined the copper environment.
  • Still unknown: The abstract does not state how many candidates were screened or list explicit limitations; full text not read.
  • AI role: DFT calculations combined with active machine learning identified the Cu-Al catalysts.

Evidence sources: https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=DOI:10.1038/s41586-020-2242-8&resultType=core&format=json

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review scope and date are not recorded; source checked does not establish full-text review.

Citations: in the top 1% of works of the same type, year and subfield (OpenAlex work metrics, 1624 citations; as of 15 Nov 2026). Attention, not a measure of correctness; not part of the score.

Editorial assessment: Impact 4/5 · Validation 3/5 · Novelty 4/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Follow-up and relationships

No follow-up recorded; this is not a claim that no follow-up exists.

Follow-up review scope and date are not recorded.

No source-supported relationships recorded.

Open the filtered catalog for charts and filters, or browse all organizations and model families.