📚 Best AI tools for Datasets & benchmarks (2026)

Reference datasets, atlases and benchmarks for AI in biology.

13 tools · 12 open source · each entry links to its website, code and paper.

Dated by release date, else first public paper, preprint or release date; undated tools listed after, A–Z.

Tahoe-100M

100M-cell single-cell drug perturbation atlas.

Datasets & benchmarksOpen source#perturbation#dataset

ProteinGym

Benchmarks for protein fitness and mutation effect prediction.

Datasets & benchmarksOpen source#benchmark

PoseBusters

Plausibility checks and benchmark for docking/structure poses.

Datasets & benchmarksOpen source#benchmark#docking

Polaris

Benchmarking platform for ML in drug discovery.

Datasets & benchmarksOpen source#benchmark#drug

Living benchmarks for single-cell analysis tasks.

Datasets & benchmarksOpen source#benchmark#single-cell

CZ CELLxGENE

Curated single-cell data portal and Census API with 100M+ cells.

Datasets & benchmarksOpen source#atlas#single-cell

ML-ready datasets and benchmarks for therapeutics.

Datasets & benchmarksOpen source#benchmark#drug

Public single-cell and spatial (Visium, Xenium) example datasets.

Datasets & benchmarksOpen source#spatial#single-cell

Critical Assessment of protein Structure Prediction community experiment.

Datasets & benchmarksOpen source#benchmark#structure

Hub listing of biology-tagged models.

Datasets & benchmarksFree (research)#hub

Data portal of the Human Cell Atlas consortium.

Datasets & benchmarksOpen source#atlas

Archive of experimentally determined 3D biomolecular structures.

Datasets & benchmarksOpen source#structure#database

Comprehensive resource of protein sequence and function.

Datasets & benchmarksOpen source#protein#database