MASS: predict the global qualities of individual protein models using random forests and novel statistical potentials

Table 3 Evaluations of our method MASS with seven top-ranking single-model methods in stage 2 for 57 targets of CASP 13 (groups are ranked by wmPMCC and best results are highlighted in bold)

Group ID	wmPMCC	Ave loss	Ave ΔGDT	MCC	ROC
ModFOLD7	0.906	0.0936	0.00041	0.72	0.94
ModFOLD7_cor	0.888	0.09313	0.00039	0.74	0.94
ModFOLD7_rank	0.839	0.05807	0.00091	0.64	0.93
FaeNNz	0.78	0.09127	0.00083	0.58	0.89
ProQ4	0.773	0.08708	0.00106	0.57	0.86
MESHI-enrich-server	0.756	0.08826	0.00087	0.52	0.88
MESHI-corr-server	0.742	0.08727	0.00088	0.57	0.88
VoroMQA-A	0.721	0.08322	0.00098	0.34	0.87
MUFold_server	0.714	0.08675	0.00095	0.6	0.89
VoroMQA-B	0.69	0.07854	0.001	0.33	0.86
MASS	0.682	0.09037	0.00106	0.54	0.85
MULTICOM-NOVEL	0.667	0.07839	0.00113	0.38	0.83
MASS2	0.652	0.09748	0.00124	0.46	0.83
Bhattacharya-SingQ	0.638	0.08676	0.00097	0.46	0.81
Bhattacharya-Server	0.601	0.11021	0.00106	0.44	0.82
PLU-AngularQA	0.57	0.13504	0.00097	0.44	0.83
PLU-TopQA	0.026	0.20285	0.00165	0.21	0.65

ISSN: 1471-2105