ChAI-Net learns a regulatory code for chromatin co-accessibility
Source check: Source checked. Not independent scientific confirmation.
Finding: Identifies distinct sequence motifs and motif arrangement/orientation patterns associated with co-accessibility at enhancers, promoters and CTCF-bound regions in mouse embryonic stem cells, including patterns missed by average-accessibility modeling. Held-out test set (MSE 0.0079; Pearson r=0.46) plus in-silico motif insertion/ablation; no direct experimental perturbation validation reported in the reviewed results.
Technology: Deep convolutional neural network predicts chromatin co-accessibility maps from DNA sequence; interpretation provides motif-level regulatory hypotheses.
Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.
- AI contribution
- Deep convolutional neural network predicts chromatin co-accessibility maps from DNA sequence; interpretation provides motif-level regulatory hypotheses.
- Test and result
- Identifies distinct sequence motifs and motif arrangement/orientation patterns associated with co-accessibility at enhancers, promoters and CTCF-bound regions in mouse embryonic stem cells, including patterns missed by average-accessibility modeling.
- Limitation / still unknown
- Held-out test set (MSE 0.0079; Pearson r=0.46) plus in-silico motif insertion/ablation; no direct experimental perturbation validation reported in the reviewed results.
Evidence sources and annotation context
Evidence context (descriptions follow the cited sources; test type is a catalog classification):
Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC13734088/
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: 22 Nov 2026. Review scope: Read the full bioRxiv v1 preprint in PMC (2026-10-31). Methods/Results show ChAI-Net is a deep CNN trained on ~3.4 million 2,048-bp windows of mouse embryonic stem-cell SAMOSA single-molecule footprint data to predict 16-bp-resolution co-accessibility maps. Held-out testing reports MSE 0.0079 and Pearson r=0.46; model interpretation and in-silico motif insertion/ablation identify sequence motifs and spacing/orientation patterns associated with co-accessibility, including patterns not recovered by average-accessibility modeling. This is a new, model-derived regulatory finding, with validation limited to held-out data and computational perturbations in one cell context (not causal wet-lab perturbation). The full text identifies Version 1 dated 2026-10-31 as containing the result. Exact DOI and title were checked and are absent from the active and archived catalog snapshots..
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 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.