Figure 7From: Predicting protein-ATP binding sites from primary sequence through fusing bi-profile sampling of multi-view features ROC curves of eight training datasets with different sizes based on 5-fold cross validation tests on ATP168. 1:1, 1:2, …, and 1:8 mean that the sizes of negative training subsets are 1 ~ 8 times respectively the sizes of corresponding positive datasets. Back to article page