Fig. 3
From: Drug-target interaction prediction using semi-bipartite graph model and deep learning
10-fold cross-validation performance evaluation of our approach compared with baseline methods in terms of AUROC and AUPR. a AUROC and b AUPR scores, in which all methods are trained and tested on balanced datasets. c AUROC and d AUPR scores in which number of negative samples was 10 times more than the number positive samples (α=10%). e AUROC and f AUPR scores in which all unknown drug-target interacting pairs are considered (α=0.18%). All results were summarized over 10 trials and expressed as mean ± SD