Comparison of In Silico Tools for Splice-Altering Variant Prediction Using Established Spliceogenic Variants: An End-User’s Point of View

Table 2

Summary of evaluation measures for 7 in silico tools only for de novo or cryptic splice sites.

In silico tools	AUC (95% CI)	Cutoff	TP	FP	TN	FN	Sensitivity	Specificity
SpliceAI	0.972 (0.937–1.008)	0.02	44	2	62	2	0.957	0.969
SpliceRover	0.924 (0.838–1.010)	0.095393	40	0	64	12	0.769	1
SPiCE	0.508 (0.457–0.559)	0.49529	0	0	64	7	0	1
MES	0.827 (0.750–0.903)	0.118125	31	0	62	15	0.674	1
NNSplice	0.818 (0.736–0.901)	0.07	28	2	60	15	0.651	0.968
SSF	0.829 (0.757–0.901)	0.0014	32	3	61	14	0.696	0.953
HSF	0.847 (0.774–0.921)	0	27	0	64	12	0.692	1

Abbreviations: AUC, area under the curve; TP, true positives; FP, false positives; TN, true negatives; FN, false negatives.