Predicting drivers of drug response from baseline omics data across breast cancer cell lines and models.

Using phosphoproteome, proteome, transcriptome, and kinase activity we are able to rank the feature importance of about 280 genes (55 of which were nominal) and use this information to create a predictive performance model of a drug. Made evident in the figures on this poster, this information has implications in drug selection as well as discovering markers for novel drugs.

DeepDyeDrop: a high-throughput microscopy platform for phenotyping the response of cancer cell lines to therapeutic agents