Fig. 1.

Validation results for ECFP4 for unprocessed, folded and filtered fingerprints (folded/filtered bit-vector size is 1024). For the random forest (RF) algorithm, 5 measures are provided for each of the 76 datasets. The difference between folded fingerprint features and filtered/unprocessed features is less distinct considering the area under the ROC curve (AUROC), and enrichment factor (EF), and more distinct considering the area under precision recall curve (AUPRC). Run-time measure the seconds to mine fragments and build a model and is highest for unprocessed features. The remaining charts for support vector machines (SVM) and naive Bayes (NB) are provided in Additional file 2