Prediction on unseen dataset without overlap

[xinghao-1210]

Hi, I would like to understand the the appropriate way to predict the response on my own dataset using chemical/drug perturbation datasets in the tutorial.

My dataset have no overlapped celltypes (or species), treatments with the perturbation studies from the tutorial. How shall I set the proper train model to predict the perturbation effects? Or whether such task is proper for implementing cpa.

Thank you