Performance of iDPath. (A) The precision–recall (PR) curves of all the models on the testing set, the values in the bracket denote the AUPRC. (B) The TN rate, false-negative rate, false-positive rate and TP rate of iDPath on the testing set. (C) The performance (NDCG@K) of all the models on the drug recommendation task on the testing set. These models are trained on the binary classification task and used to generate the repurposing probabilities of all the drugs on different diseases in the testing set. (D) The performance of iDPath with different biological network layers. Here GRN–PPI–PCI–CCI denotes the multilayer biological network generated by these four networks, GRN denotes using gene regulatory network alone, and the same goes for PPI and PCI. The K values in c and d denote the top |$K$| drugs used to compute for |$NDCG$|⁠.