MIA: Lotfollahi, Deep perturbation modelling; Fischer, Learning cell communication

Models, Inference and Algorithms Broad Institute of MIT and Harvard September 29, 2021 Mohammad Lotfollahi Technical University of Munich David Fischer Institute of Computational Biology, Helmholtz Zentrum München Deep interpretable perturbation modeling in single cell genomics; Learning cell communication from spatial graphs of cells Recent advances in multiplexing singlecell transcriptomics across experiments are enabling the highthroughput study of drug and genetic perturbations. However, an exhaustive exploration of the combinatorial perturbation space is experimentally unfeasible, so computational methods are needed to predict, interpret and prioritize perturbations. Here, we present the Compositional Perturbation Autoencoder (CPA), which combines the interpretability of linear models with the flexibility of deeplearning approaches for singlecell response modeling. CPA encodes and learns transcriptional drug response acr