Fig. 1From: DDI-PULearn: a positive-unlabeled learning method for large-scale prediction of drug-drug interactionsF1-scores of DDI-PULearn with different PCNs. The x-axis is the PCA component number and the y-axis is the F1-score. Panel (a) shows the F1-scores for PCN between 1 and 2000, and Panel (b) is an amplification of the range [20,150] (amplification ratio = 5)Back to article page