Table 3.

The predictive performance of TRGP, LDA, SVM, and random forest classifier.

Classifier	Data sets	Number of gene pairs	Accuracy	Sensitivity	Specificity	Precision	Balanced accuracy	AUROC
LDA	Discovery set	60	0.978	0.993	0.949	0.973	0.971	0.980
	Validation set-1		0.898	0.882	0.904	0.763	0.893	0.860
	Validation set-2		0.589	0.543	0.695	0.802	0.619	0.601
SVM	Discovery set	12	0.955	0.979	0.910	0.953	0.945	0.956
	Validation set-1		0.817	0.980	0.760	0.588	0.870	0.790
	Validation set-2		0.698	0.699	0.695	0.839	0.697	0.671
Random forest	Discovery set	56	1.000	1.000	1.000	1.000	1.000	1.000
	Validation set-1		0.868	0.961	0.836	0.671	0.898	0.828
	Validation set-2		0.788	0.866	0.610	0.835	0.738	0.751
TRGP	Discovery set	7	0.830	0.936	0.774	0.689	0.855	0.947
	Validation set-1		0.934	0.959	0.863	0.952	0.911	0.955
	Validation set-2		0.703	0.797	0.662	0.508	0.729	0.796