Figure 4.

DL models for predicting drug response and synergy. (A) Models for predicting drug sensitivity of a single drug and simultaneous comparisons of multiple drugs. The former model is composed of subnetworks learning from a sample’s genomics profile and chemical descriptor/fingerprint of a drug and yields a drug response score. The latter takes only genomics profiles of samples as input. Each node at the output layer predicts response to a drug. (B) Model for predicting drug synergy. The model learns from genomics of a sample and descriptors/fingerprints of two drugs and outputs a synergy score.