theislab / 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.
BSD 3-Clause "New" or "Revised" License
83 stars 17 forks source link

which version ? The theislab version or facebook version? #9

Open bitcometz opened 2 years ago

bitcometz commented 2 years ago

Hello,

I found two version of CPA:

(a) the theislab version, which is built on scvi-tools. (b) the facebook version, which is built on naive pytorch.

I want to reproduce the analysis results here from cpa-reproducibility

I found the results generated by the facebook version is quite different from the paper, so I should use the theislab version ?

Thanks !!!