Single‑cell method ecosystem

Multi‑Task Omics

An ecosystem of methods for multimodal, multibatch multi‑condition single‑cell analysis.

$ pip install m3‑sc multibench matilda‑sc

GitHub Read the papers ↗ 3 packages · 11 tasks

Packages

M3 Factorised, condition‑aware embeddings — multimodal integration with patient‑level inference. IntegrationPatient predictionAttribution scMultiBench Run & score 40+ integration methods with scIB metrics — rankings, bubble tables, an interactive explorer. IntegrationBatch correctionBenchmarking 🚧 In preparation — coming soon. Matilda One network for many tasks — trained once, reused across the analysis pipeline. ClassificationDimension reductionFeature selectionSimulation

References

Condition-aware deep factor learning with M3 enables integration, patient-level inference, and multi-resolution interpretation in multimodal single-cell data Chunlei Liu, Sichang Ding, Shila Ghazanfar, Pengyi Yang

Multitask benchmarking of single-cell multimodal omics integration methods Chunlei Liu, Sichang Ding, Hani Jieun Kim, Siqu Long, Di Xiao, Shila Ghazanfar, Pengyi Yang Nature Methods · 22, 2449–2460 (2025) Multi-task learning from multimodal single-cell omics with Matilda Chunlei Liu, Hao Huang, Pengyi Yang Nucleic Acids Research · 51(8), e45 (2023)

Independent packages — install only what you need. Which tool do I need? →