Fig. 3.

Performance of the different systems trained on the different datasets evaluated using accuracy (ACC), Matthews Correlation Coefficient (MCC), F1, recall and precision as defined in the formulas (Supplementary Tables S5, S6). GIS-MaxEnt trained on the six different training sets: GO-BP, GO-CC, GO-MF, GO, GO-IP, SVM trained on the same six training sets: SVM-GO-BP, SVM-GO-CC, SVM-GO-MF; SVM-GO, SVM-IP, SVM-GO-IP; GIS-MaxEnt Ensemble (GME) and Multiple Kernel Learning (MKL) which were trained on all the data