MultiPert Predicts Single-Cell Multi-Omics Perturbation Responses
Researchers (Zhao et al.) on March 11, 2026 published MultiPert, a deep-learning framework for predicting single-cell multi-omics perturbation responses using modality-specific encoders, dual-attention fusion, and adversarial cross-modal alignment. Benchmarks on THP-1 and kidney datasets show improved accuracy for perturbed gene expression and protein abundance, generalization to unseen perturbations, and interpretable regulatory insights, offering a foundation for pathogenesis studies and drug discovery.
Key Points
- 1Introduces MultiPert, a model using modality-specific encoders and dual-attention for perturbation prediction.
- 2Achieves superior accuracy predicting perturbed gene expression and protein abundance on THP-1 and kidney datasets.
- 3Enables generalization to unseen perturbations and interpretable regulatory insights for drug discovery.
Scoring Rationale
Strong, peer-reviewed multi-omics prediction method with solid benchmarks; limited generalizability beyond tested THP-1 and kidney datasets.
Sources
Public references used for this report.
Practice with real Ad Tech data
90 SQL & Python problems · 15 industry datasets
250 free problems · No credit card
See all Ad Tech problems
