DeepMILO: a deep learning approach to predict the impact of non-coding sequence variants on 3D chromatin structure
Source check: Source checked. Not independent scientific confirmation.
Finding: This method paper reports a specific cancer-genomics result (recurrent predicted loop disruption and association with BCL2/MYC expression), supported by learned sequence modeling and genomic expression analysis. Keep associations and predictions provisional. Methods/Results: CNN/RNN learn DNA sequence features for CTCF anchor/orientation and loop predictions. Test AUPRC varies by negative set (0.684 hardest); application to WGS of 1,834 patients across 12 cancer types predicts 672 insulator loops disrupted in ≥10% of patients; variants at anchors associated with BCL2/MYC upregulation in malignant lymphoma. Association/predictions computational, with validation on published loop datasets/deletion examples; does not prove causal mechanism.
Technology: CNN/RNN learn DNA sequence features for CTCF anchor/orientation and loop predictions. Test AUPRC varies by negative set (0.684 hardest); application to WGS of 1,834 patients across 12 cancer types predicts 672 insulator loops disrupted in ≥10% of patients; variants at anchors associated with BCL2/MYC upregulation in malignant lymphoma. Association/predictions computational, with validation on published loop datasets/deletion examples; does not prove causal mechanism.
Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.
- AI contribution
- CNN/RNN learn DNA sequence features for CTCF anchor/orientation and loop predictions. Test AUPRC varies by negative set (0.684 hardest); application to WGS of 1,834 patients across 12 cancer types predicts 672 insulator loops disrupted in ≥10% of patients; variants at anchors associated with BCL2/MYC upregulation in malignant lymphoma. Association/predictions computational, with validation on published loop datasets/deletion examples; does not prove causal mechanism.
- Test and result
- CNN/RNN learn DNA sequence features for CTCF anchor/orientation and loop predictions. Test AUPRC varies by negative set (0.684 hardest); application to WGS of 1,834 patients across 12 cancer types predicts 672 insulator loops disrupted in ≥10% of patients; variants at anchors associated with BCL2/MYC upregulation in malignant lymphoma. Association/predictions computational, with validation on published loop datasets/deletion examples; does not prove causal mechanism.
- Limitation / still unknown
- Methods/Results: CNN/RNN learn DNA sequence features for CTCF anchor/orientation and loop predictions. Test AUPRC varies by negative set (0.684 hardest); application to WGS of 1,834 patients across 12 cancer types predicts 672 insulator loops disrupted in ≥10% of patients; variants at anchors associated with BCL2/MYC upregulation in malignant lymphoma. Association/predictions computational, with validation on published loop datasets/deletion examples; does not prove causal mechanism. The source does not establish benefits beyond the reported evaluation.
Evidence sources and annotation context
Evidence context (descriptions follow the cited sources; test type is a catalog classification):
Evidence sources: https://link.springer.com/article/10.1186/s13059-020-01987-4
Source checked · editorial details
The source page was checked against this entry. Independent confirmation is not assessed by this label.
Source type: other. Paper: not linked in this catalog; this does not mean no paper exists.
Review date: 22 Nov 2026. Review scope: Primary result-bearing material for the stated method, result and evaluation, with bibliographic version information..
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 2/5 · Validation 2/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.