Fig. 8
From: How to approach machine learning-based prediction of drug/compound–target interactions
Regression-based test performance results of protein family-specific PCM models (each using a different representation type as an input feature vector) for random-split, dissimilar-compound-split, and fully-dissimilar-split datasets based on (a) median corrected RMSE, and (b) Spearman correlation scores. The models are ranked according to decreasing performance on the fully-dissimilar-split dataset (i.e., the large-scale analysis)