Table 2.

Comparisons of C. elegans.

Tools	Sensitivity	Specificity	Accuracy	F-measure	MCC	Kappa
CPC.web	0.9990	0.9895	0.9942	0.9942	0.9885	0.9884
CPAT.web	0.9910	0.9187	0.9548	0.9564	0.9120	0.9097
CNCI	0.9938	0.7744	0.8841	0.8955	0.7873	0.7681
PLEK	0.9987	0.4526	0.7256	0.7845	0.5387	0.4513
Lncident∗	1	0.9545	0.9772	0.9778	0.9555	0.9545
CPAT.train∗∗	0.9995	0.9950	0.9972	0.9973	0.9945	0.9945
PLEK.train∗∗	0.9795	0.9950	0.9872	0.9872	0.9746	0.9745
Lncident.train∗∗	0.9950	0.9975	0.9962	0.9962	0.9925	0.9925

For tools with default models, Lncident presented the best result among the alignment-free methods. Both Lncident and CPAT outperformed CPC by utilizing new-trained model. ^∗Lncident with model trained on human. ^∗∗The suffix of “train” means the tools with model trained on C. elegans.