arcinstitute/Perturb-Sapiens

Perturb Sapiens: A Human Whole-Organism Atlas of Perturbed Cells

Dataset Description

Perturb Sapiens is an evolving database of AI-predicted single-cell perturbation responses, representing the first human whole-organism atlas of perturbed cells.

Perturb Sapiens is generated using the post-trained Stack model (Stack-Large-Aligned), an in-context learning foundation model for single-cell biology.

Data Sources:

• Prompt Data: Parse/OpenProblems PBMC perturbation data
• Query Data: Tabula Sapiens

Key Statistics (v1)

Metric	Value
Total Cells	103+ million
Human Tissues	28
Cell Classes	40
Drug Conditions	111 (control: Dimethyl sulfoxide)
Cytokine Conditions	90 (control: PBS)

Dataset Structure

PerturbSapiensv1/
├── drug/
│   ├── [drug_name].h5ad
│   └── Dimethyl sulfoxide.h5ad  # Control
└── cytokine/
    ├── [cytokine_name].h5ad
    └── PBS.h5ad                  # Control

Quick Start

import scanpy as sc

# Load a perturbation condition (e.g., ADSF cytokine)
adata = sc.read_h5ad('PerturbSapiensv1/cytokine/ADSF.h5ad')

# Preprocessing (aligned with post-training settings)
sc.experimental.pp.highly_variable_genes(adata, n_top_genes=4000)
sc.pp.normalize_total(adata)
sc.pp.log1p(adata)

# Subset to high quality cells
adata = adata[adata.obs['gen_logit']<2.5].copy()

# Subset to highly variable genes
adata = adata[:, adata.var['highly_variable']].copy()

# Ready for downstream analysis!

Intended Uses

• Drug response prediction: Explore how drugs affect gene expression across tissues
• Cytokine signaling studies: Understand tissue-specific responses to immune signals
• Hypothesis generation: Identify potential therapeutic targets

Limitations

• Predictions are generated by a foundation model and should be validated experimentally
• Potentially less precision in predicting effects of weak perturbations

Citation

If you use Perturb Sapiens in your research, please cite:

• Dong et al., 2026: Stack: In-context modeling of single-cell biology. Preprint. Paper link

Links

• 🔗 GitHub: ArcInstitute/stack
• 🤗 Model: Stack-Large-Aligned

License

This dataset is released under the CC BY 4.0 License.

Contact

For questions or issues, please open an issue on the GitHub repository.