Graph neural networks trained on a 2,352-compound screen find three selective senolytic compounds
Source check: Needs source check. Not independent scientific confirmation.
Finding: Wong et al. screened 2,352 compounds for senolytic activity in etoposide-induced senescence, trained graph neural networks on the results and scored more than 800,000 molecules. Three drug-like hits selectively killed senescent cells across senescence models with selectivity comparable to ABT-737; docking and TR-FRET suggest partial Bcl-2 inhibition. One compound, BRD-K56819078, reduced senescent-cell burden and senescence-associated gene expression in kidneys of aged mice. Evidence is from the abstract only; effect sizes, mouse numbers and the absence of efficacy or safety data in humans are not assessed here.
Technology: Chemprop message-passing graph neural networks
Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.
- AI contribution
- Graph neural networks trained on the experimental screen ranked more than 800,000 molecules, enriching for structurally diverse senolytic hits.
- Test and result
- Screening hits were tested in cell models of senescence, mechanism was probed by docking and TR-FRET, and one compound was tested in aged mice (per abstract).
- Limitation / still unknown
- Independent replication, human data and full Methods were not assessed.
Evidence sources and annotation context
Evidence context (descriptions follow the cited sources; test type is a catalog classification):
Evidence sources: https://europepmc.org/article/MED/37142829
Needs source check · editorial details
The source has not been fully checked or contains conflicting information. 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: abstract and bibliographic metadata (Europe PMC) only; Methods and full text not assessed.
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.
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.