Evaluation of in silico pathogenicity prediction tools for the classification of small in-frame indels

BMC Medical Genomics

Table 3 Performance metrics for all indel pathogenicity prediction tools tested

Tool	TP	FP	TN	FN	Total (%)	Sens	Spec	LR+	LR−	PPV	NPV	AUC	MCC
All variants (1740 pathogenic; 2224 benign)
CADD	852	176	2047	886	3961 (99.9)	0.49	0.92	6.19^$	0.55	0.83	0.70	0.86	0.47
CAPICE	1611	715	1500	129	3955 (99.7)	0.93	0.68	2.87⁺	0.11⁺	0.69	0.92	0.91	0.61
FATHMM-Indel	1626	572	1622	111	3931 (99.2)	0.94	0.74	3.59⁺	0.09⁺	0.74	0.94	0.91	0.68
MutPred-Indel	516	72	2116	1211	3915 (98.8)	0.30	0.97	9.08^$	0.73	0.88	0.64	0.81	0.37
MutationTaster2021	1675	99	1976	36	3786 (95.5)	0.98	0.95	20.52^$	0.02^$	0.94	0.98	–	0.93
PROVEAN	1482	581	1470	82	3615 (91.2)	0.95	0.72	3.35⁺	0.07^$	0.72	0.95	0.93	0.66
SIFT-Indel	1406	827	1287	313	3833 (96.7)	0.82	0.61	2.09	0.30⁺	0.63	0.80	–	0.43
VEST-indel	1560	385	1758	179	3882 (98.0)	0.90	0.82	4.99⁺	0.13⁺	0.80	0.91	0.93	0.71
VVP	1720	728	1481	20	3949 (99.6)	0.99	0.67	3.00⁺	0.02^$	0.70	0.99	0.87	0.67
DDD subset (70 pathogenic; 81 benign)
CADD	45	16	65	25	151 (100)	0.64	0.80	3.25⁺	0.45	0.74	0.72	0.78	0.45
CAPICE	64	44	37	6	151 (100)	0.91	0.46	1.68	0.19⁺	0.59	0.86	0.82	0.41
FATHMM-indel	66	38	43	4	151 (100)	0.94	0.53	2.01	0.11⁺	0.63	0.91	0.74	0.51
MutPred-Indel	17	16	64	53	150 (99.3)	0.24	0.80	1.21	0.95	0.52	0.55	0.64	0.05
MutationTaster2021	50	19	61	19	149 (98.7)	0.72	0.76	3.05⁺	0.36⁺	0.72	0.76	–	0.49
PROVEAN	60	32	45	6	143 (94.7)	0.91	0.58	2.19	0.16⁺	0.65	0.88	0.86	0.51
SIFT-indel	60	39	41	10	150 (99.3)	0.86	0.51	1.76	0.28⁺	0.61	0.80	–	0.39
VEST-indel	62	29	51	8	150 (99.3)	0.89	0.64	2.44⁺	0.18⁺	0.68	0.86	0.87	0.53
VVP	68	70	11	2	151 (100)	0.97	0.14	1.12	0.21⁺	0.49	0.85	0.64	0.19

Findings from the entire dataset are included in the top table, and just the novel (DDD-only) subset in the bottom table. Relative strength of likelihood ratios for application of ‘strong’ or ‘moderate’ evidence under the ACMG/ACGS variant classification criteria [33] are denoted as: ^$High relative strength, ⁺Medium relative strength. Additional file 1: Fig. S1 shows ROC-AUC curves
TP True positive, FP False positive, TN True negative, FN False negative, LR + positive likelihood ratio, LR− negative likelihood ratio, PPV Positive predictive value, NPV negative predictive value, AUC Area under the curve, MCC Matthews correlation coefficient

ISSN: 1755-8794