Table 3. Method comparison on Testing-A and Testing-B of hg19.

	Testing-A					Testing-B
	SES(%)	SPC(%)	ACC(%)	MCC(%)	AUC(%)	SES(%)	SPC(%)	ACC(%)	MCC(%)	AUC(%)
CPC	97.62	67.23	82.425	68.069	94.55	97.328	69.262	82.831	68.898	94.78
CPAT	85.28	94.60	89.94	80.23	95.17	83.941	95.223	89.768	79.896	95.14
lncRScan-SVM	89.20	93.88	91.54	83.17	96.39	88.215	94.479	91.45	82.985	96.39
iSeeRNA	87.97	92.32	90.13	80.36	95.33	87.04	92.965	90.082	80.238	95.28
iSeeRNA2	90.02	92.409	91.205	82.44	96.23	89.103	92.885	91.045	82.106	96.18
RNAcon	69.11	84.53	76.82	54.29	86.11	68.039	85.454	77.034	54.475	86.1

CPC and CPAT were run by submitting the GTF files of Testing-A and Testing-B through their web interfaces. The lncRScan-SVM predictor was run after feature scaling. Besides, we tested the default iSeeRNA predictor and iSeeRNA2, an iSeeRNA model re-trained on Training-A, as well as RNAcon with a parameter T equals 0. The biggest value in each column is in bold font while the smallest one is underlined.