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.
In the tutorial, you mentioned that "if we want to use CPA to predict X+Y when either X, Y, or both are unseen" in predicting Crispr perturbation, we need to use external gene embedding, but I cannot figure out how to provide this to CPA although I follows the drug example you referred? So, could you give a more detailed example about providing gene embedding to CPA?
Thanks for the great jobs!
In the tutorial, you mentioned that "if we want to use CPA to predict X+Y when either X, Y, or both are unseen" in predicting Crispr perturbation, we need to use external gene embedding, but I cannot figure out how to provide this to CPA although I follows the drug example you referred? So, could you give a more detailed example about providing gene embedding to CPA?
Thanks for your reply.