Fig. 5From: SWnet: a deep learning model for drug response prediction from cancer genomic signatures and compound chemical structuresPhysical interactions often occur among drug target proteins and proteins encoded by genes with the highest weight (\(weight=1\)). a Five BRAF inhibitors were included in the training data. Genes with \(weight=1\) were extracted from the gene weight layer after model training, and their interactions with BRAF were analyzed in the String database. Connections within 2 degrees from BRAF were shown, grouped by drug. b Three BCL2 inhibitors was included in the training data. Analysis procedure was the same as above. The network showed connections within 2 degrees from BCL2Back to article page