Table 3.

Performance of funtrp models for training and independent test sets

		ntModel	funtrpModel	random	PTEN	TPMT	HSP90
Precision	Neutral	0.94	0.90	0.53	0.84	0.91	0.90
	Rheostat	-	0.73	0.28	0.70	0.64	0.34
	Toggle	0.88	0.88	0.20	0.87	0.88	0.73
Recall	Neutral	0.95	0.91	0.32	0.87	0.90	0.73
	Rheostat	-	0.77	0.36	0.78	0.71	0.62
	Toggle	0.85	0.80	0.35	0.73	0.83	0.67
Overall Accuracy		92.3%	85.1%	55.5%	79.3%	84.1%	70%

Model performance was evaluated separately for training sets using Leave-One-Out-Cross-Validation (LOO-CV) and three independent test sets not seen before. Performance measures were calculated using Equation (3).