Research map / Label-efficient and robust learning / Recognition and long-tail
scDEBART: Predicting in silico Single-Cell Perturbation Responses via Large-Scale Differential Expression Learning
Jieun Sung, Wankyu Kim
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 |
|---|---|---|---|
| add pretraining on 6.28M differential expression profiles (ablation) | scDEBART · Norman K562, top-50 | enrichment factor: 12.693 -> 14.901 | Improved |
| add pretraining (ablation) | scDEBART · Norman K562, top-50 | cosine similarity: 0.825 -> 0.852 | No clear effect |
| add perturbation-response cell filtering in fine-tuning (ablation) | scDEBART · Norman K562, top-50 | enrichment factor: 14.349 -> 14.901 | No clear effect |
| add cell filtering (ablation) | scDEBART · Norman K562, top-50 | RMSE: 0.463 -> 0.494 | Mixed |
| increase pretraining corpus from 50% to 100% (ablation) | scDEBART · Norman K562, top-50 | enrichment factor: 12.51 -> 14.33 | Improved |
| increase pretraining corpus size (ablation) | scDEBART · Norman K562 | cosine similarity: 0.772, 0.779, 0.775 (25%, 50%, 100%) | No clear effect |
| inject uncertainty-proportional noise into scVI logFC targets at fine-tuning (ablation) | scDEBART · Norman K562, top-50 | enrichment factor: 15.992 -> 15.453 (gamma 0.5), 15.370 (gamma 1.0) | Hurt |
| apply scVI denoising to pretrained scGPT inputs | scGPT · Perturb-seq top-50 cosine similarity | cosine similarity: 0.45 -> 0.70 | Improved |
| apply scVI denoising to non-pretrained linear regression and MLP | LR, MLP · Perturb-seq top-50 cosine similarity | cosine similarity: 0.62 -> -0.27 (LR), 0.63 -> -0.27 (MLP) | Hurt |
| use simple MLP/LR instead of GEARS/scGPT on raw counts | count-based baselines · Perturb-seq top-50 cosine similarity | cosine similarity: 0.14 (GEARS), 0.45 (scGPT) -> 0.63 (MLP), 0.62 (LR) | Improved |
| restrict pretraining to high-detection, high-confidence DE genes | scDEBART · top-50 predicted genes | median non-zero proportion: baselines 0.09-0.28 -> 0.60 (perturbed), 0.73 (control) | Improved |
| fine-tune scDEBART on genetic perturbations then apply to drug perturbations without extra training | scDEBART · SCIPLEX K562, top-20 DEGs | cosine similarity: 0.04 (10 nM) -> 0.30 (10000 nM) | Mixed |
Closest papers by meaning
- PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (NeurIPS 2025 · similarity 0.88)
- PertEval-scFM: Benchmarking Single-Cell Foundation Models for Perturbation Effect Prediction (ICML 2025 · similarity 0.87)
- A benchmark for prediction of transcriptomic responses to chemical perturbations across cell types (NeurIPS 2024 · similarity 0.87)
- MapPFN: Learning Causal Perturbation Maps in Context (NeurIPS 2026 · similarity 0.86)
- PRESCRIBE: Predicting Single-Cell Responses with Bayesian Estimation (NeurIPS 2025 · similarity 0.86)
- Chem-PerturBridge: a harmonized compendium of small molecule perturbation transcriptomic effects (NeurIPS 2026 · similarity 0.86)
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.
