Fig. 1
From: Predicting circRNA-drug sensitivity associations via graph attention auto-encoder
The flowchart of GATECDA. (1) We build a comprehensive similarity matrix CS for circRNAs and a comprehensive similarity matrix DS for drugs, respectively. (2) Two thresholds are utilized to binarize the corresponding comprehensive similarity matrices CS and DS. (3) GATE is employed to extract the representations of circRNAs and drugs respectively. (4). The representations of circRNAs and drugs are combined and fed into a fully connected neural network to predict the associations of each pair of circRNA and drug sensitivity