facebookresearch / CPA

The Compositional Perturbation Autoencoder (CPA) is a deep generative framework to learn effects of perturbations at the single-cell level. CPA performs OOD predictions of unseen combinations of drugs, learns interpretable embeddings, estimates dose-response curves, and provides uncertainty estimates.
MIT License
175 stars 48 forks source link

Merging dev branch with changes to API #7

Closed cdedonno closed 2 years ago

cdedonno commented 2 years ago

Changes include: