Open Problems is a framework for machine learning scientists, computational biologists, and single-cell data analysts to benchmark and compare single-cell omics methods on a variety of tasks. It combines a permissively licensed GitHub repository with community-defined tasks, an automated benchmarking workflow, and a website to explore the results. 9 million cells are represented in Open Problems.
OpenProblems offers a collection of common datasets in single-cell analysis that can be used to test components and run the benchmarking tasks.
OpenProblems contains a shared core infrastructure, including a common dataset repository, shared compute infrastructure, and a unified platform for exploration of benchmarking results.
Benchmarks are shaped and enhanced by community contributions.
- 10.1038/s41587-025-02694-w: Luecken, M. D. et al. Defining and benchmarking open problems in single-cell analysis. Nat. Biotechnol. 43, 1035–1040 (2025).