Research map / Probabilistic and scientific modeling / Causal discovery
Interpretable Neural ODEs for Gene Regulatory Network Discovery under Perturbations
Zaikang Lin, Sei Chang, Aaron Zweig, Minseo Kang, Fabian J Theis, Elham Azizi, David A. Knowles
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| learn tunable overexpression strength (PerturbODE*) instead of fixed (ablation) | PerturbODE (imperfect interv) · TF Atlas human reference GRNs | recall: 0.3659 -> 0.4976 | Improved |
| use imperfect rather than perfect intervention modeling (ablation) | PerturbODE · TF Atlas human reference GRNs | recall: 0.3561 -> 0.3659 | Improved |
| increase number of modules from 10 to 350 (ablation) | PerturbODE · AUPRC, TF Atlas vs ChIP-Atlas (223 genes) | AUPRC: 0.017528 -> 0.022517 | Improved |
| increase number of modules from 350 to 400 (ablation) | PerturbODE · AUPRC, TF Atlas vs ChIP-Atlas (223 genes) | AUPRC: 0.022517 -> 0.021096 | No clear effect |
| switch ODE solver from dopri5 to euler (ablation) | PerturbODE · AUPRC, ChIP-Atlas (223 genes) | AUPRC: 0.017 -> 0.018 | No clear effect |
| lower learning rate from 1e-2 to 1e-4 (ablation) | PerturbODE · AUPRC, TF Atlas | AUPRC: 0.021 -> 0.022 | No clear effect |
| raise learning rate to 1e-1 or 1 (ablation) | PerturbODE · AUPRC, TF Atlas | AUPRC: 0.022 -> OOM | Hurt |
| use PerturbODE instead of GENIE3 at medium threshold | GENIE3 · ChIP-Atlas reference GRN | FDR (%): 54.1 -> 52.5 | Improved |
| use PerturbODE instead of GENIE3 at low threshold | GENIE3 · ChIP-Atlas reference GRN | FDR (%): 54.1 -> 55.4 | Hurt |
| use PerturbODE instead of GEARS for held-out intervention prediction | GEARS · TF Atlas held-out interventions | W2 distance: 65 -> 84 | Hurt |
| use PerturbODE instead of scGPT for held-out intervention prediction | scGPT · TF Atlas held-out interventions | W2 distance: 91 -> 84 | Improved |
| use PerturbODE instead of NO-TEARS for held-out intervention prediction | NO-TEARS · TF Atlas held-out interventions | W2 distance: 396 -> 84 | Improved |
Closest papers by meaning
- Identifying biological perturbation targets through causal differential networks (ICML 2025 · similarity 0.87)
- MapPFN: Learning Causal Perturbation Maps in Context (NeurIPS 2026 · similarity 0.86)
- Contextualizing biological perturbation experiments through language (ICLR 2025 · similarity 0.85)
- KINDLE: Knowledge-Guided Distillation for Prior-Free Gene Regulatory Network Inference (NeurIPS 2025 · similarity 0.85)
- PerturbReason: A Knowledge-Grounded Benchmark and Framework for Cell-State–Conditioned Mechanistic Reasoning of Perturbation Effects (NeurIPS 2026 · similarity 0.85)
- Interpretable Causal Representation Learning for Biological Data in the Pathway Space (ICLR 2025 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
