DrVAEN predicts drug response based on baseline transcriptome data that can be obtained using cancer tissues, organoid, or cell lines. We provide well-trained models for 24 anti-cancer compounds from CCLE and 251 compounds from GDSC. A full list of compounds is available here.

We provide four prediction models (please see details in Tutorial)depending on the normalization method (Rank or ZS) and the activation functions in the VAE model (Sigmoid or ReLU), with Rank+Sigmoid as the default and suggestive one.

DrVAEN allows users to run the prediction for up to 50 drugs per job. Each job may take 1-10 minutes to finish. You may retrieve the results anytime using the Job Identifier (e.g., JOB1234_1234567890). Please keep a record of the Job Identifier for future reference.

Input data format: gene expression data with rows as genes (with gene name) and columns as samples (with header line as the sample name/ID). Example datasets can be found here.