Performance of the best RP classification models for the training set and test set using different combinational of molecular properties and fingerprints.

a<p>MP represents thirteen molecular properties. The best tree depth is 8 for RP models (MP, MP+EPFP_4 and MP+LPFP_6), 10 for RP models (MP+ECFP_6 and MP+LPFP_4), 11 for RP models (MP+FCFP_4, MP+FCFP_6, MP+LCFP_4, MP+LCFP_6 and MP+FPFP_6) and 12 for RP model (MP+FPFP_4). The detailed results can be found in <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0095221#pone.0095221.s004" target="_blank">Table S1</a> and <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0095221#pone.0095221.s005" target="_blank">Table S2</a> in Supporting Information.