Table 2.

Performance of in silico prediction models with optimized thresholds for classification of BRCA1 and BRCA2 missense variants

Gene	Model	Optimal Threshold	AUC (95%CI)	FN / FP / TP / TN	MCC
BRCA1	NVM-Validation	≥9	0.94 (0.897–0.983)	5 / 8 / 25 / 79	0.719
	Vest3RankScore	≥0.85546	0.9 (0.849–0.95)	8 / 24 / 51 / 153	0.678
	Vest3Score	≥0.868	0.9 (0.849–0.95)	8 / 24 / 51 / 153	0.678
	RF	≥0.298	0.92 (0.879–0.96)	8 / 26 / 51 / 151	0.663
	AlignGVGDPrior	≥0.29	0.88 (0.829–0.931)	7 / 37 / 54 / 150	0.614
	PERCHnoMAF	≥0.206316	0.87 (0.814–0.924)	10 / 31 / 51 / 156	0.614
	PERCH	≥0.239853	0.87 (0.819–0.922)	10 / 32 / 51 / 155	0.607
	Polyphen2HvarRankScore	≥0.91584	0.89 (0.845–0.93)	11 / 30 / 48 / 147	0.593
	Polyphen2HvarScore	≥0.999	0.89 (0.845–0.93)	11 / 30 / 48 / 147	0.593
	MetaSVMRankScore	≥0.9083	0.89 (0.844–0.928)	11 / 34 / 48 / 143	0.565
BRCA2	NVM-Validation	≥4	0.89 (0.826–0.963)	6 / 9 / 29 / 59	0.683
	PERCH	≥0.295957	0.89 (0.847–0.939)	11 / 21 / 60 / 115	0.672
	PERCHnoMAF	≥0.272149	0.88 (0.832–0.929)	12 / 23 / 59 / 113	0.642
	RFModel	≥0.371	0.9 (0.843–0.947)	12 / 24 / 59 / 111	0.633
	MetaSVMRankScore	≥0.93181	0.87 (0.824–0.923)	15 / 29 / 56 / 107	0.555
	MetaSVMScore	≥0.7002	0.87 (0.824–0.923)	15 / 29 / 56 / 107	0.555
	MetaLRRankScore	≥0.92107	0.87 (0.823–0.922)	16 / 30 / 55 / 106	0.535
	MetaLRScore	≥0.7679	0.87 (0.823–0.922)	16 / 30 / 55 / 106	0.535
	Vest3RankScore	≥0.79963	0.83 (0.776–0.893)	16 / 30 / 55 / 106	0.535
	Vest3Score	≥0.811	0.83 (0.776–0.893)	16 / 30 / 55 / 106	0.535

FN: False negative; FP: False positive; TP: True positive; TN: True negative

AUC: Area under the curve from Receiver Operator characteristic analysis

MCC: Matthew Correlation Coefficient