Fig. 5
From: How to approach machine learning-based prediction of drug/compound–target interactions
t-SNE projections of train-test samples (i.e., compound-protein pairs) of transferase and ion channel families for k-sep_pssm and unirep1900 representations on; (a) the random-split, (b) dissimilar-compound-split, and (c) the fully-dissimilar-split dataset