Table 1. Tools performance on the GENCODE human data sets. Bold-underlined values correspond to the highest values of each metrics. CPAT_train and PLEK_train correspond to program versions trained with the HL data set. Programs are sorted by MCC values.

HUMAN data set	Program	Sensitivity	Specificity	Precision	Accuracy	F-score	MCC
	FEELnc	0.923	0.915	0.916	0.919	0.919	0.838
	CPAT	0.899	0.924	0.922	0.912	0.910	0.823
	CPAT_train	0.920	0.901	0.903	0.910	0.911	0.821
	CNCI	0.829	0.979	0.975	0.904	0.896	0.817
	PLEK	0.732	0.985	0.981	0.858	0.838	0.741
	PhyloCSF	0.906	0.802	0.820	0.854	0.861	0.712
	PLEK_train	0.582	0.960	0.936	0.770	0.718	0.584
	CPC	0.699	0.739	0.728	0.719	0.713	0.438