. 2019 Mar;21(2):318–329. doi: 10.1016/j.jmoldx.2018.10.009

Table 1.

Summary of the Data Sets

Source	Variant type	Samples	Unique variants	Variant calls	TPs	FPs	FDR, %	Total calls	Total FPs	FP sensitivity, %	CI lower bound, %
This study: Lab 1	SNVs	GIAB	27,202	136,146	135,945	201	0.15
	SNVs	Patients^∗^†	2840	3699	3689	10	0.27	139,845	211	100	98.9
	Indels	GIAB	3715	15,574	14,594	980	6.29
	Indels	Patients^∗^†	1749	2274	2262	12	0.53	17,848	992	100	99.8
This study: Lab 2	SNVs	GIAB	5816	29,148	29,110	38	0.13
	SNVs	Patients^∗^‡	4359	4934	4804	130	2.63	34,082	168	100	98.5
	Indels	GIAB	1185	3617	3343	274	7.58
	Indels	Patients^∗	267	389	372	17	4.37	4006	291	100	99.1
Strom et al¹⁰	SNVs	Patients^§^¶	-	108	107	1	0.93	108	1	100	5.1
Baudhuin et al¹¹	SNVs	Patients^‖	380	797	797	0	0
	SNVs	1KG^‖^∗∗	736	736	736	0	0	1533	0	N/A	N/A
	Indels	Patients^‖^††	63	122	122	0	0
	Indels	1KG^‖^∗∗	26	26	26	0	0	148	0	N/A	N/A
Mu et al¹²	SNVs	Patients^¶	-	6912	6818	94	1.36	6912	94	100	97.4
Mu et al¹²	Indels	Patients^¶	-	933	928	5	0.54	933	5	100	62.1
van den Akker et al¹⁴	SNVs	Patients^¶^‡‡	3044	5829	5524	305	5.23	5829	305	100	99.2
van den Akker et al¹⁴	Indels	Patients^¶^‡‡	526	1350	1142	208	15.41	1350	208	100	98.8

1KG, 1000 Genomes Project; FDR, false discovery rate [calculated as FPs/(FPs + TPs)]; FP, false positive; FP sensitivity, the fraction of FPs captured using the study's proposed criteria; GIAB, Genome in a Bottle Consortium; Indel, insertion or deletion; SNV, single-nucleotide variant; unique variant, a particular alteration which may be present in one or more individuals; TP, true positive.

^∗

Clinical and GIAB data were combined in the final analysis. GIAB data included on- and off-target calls.

^†

Manual review removed certain FPs (particularly indels) and thus reduced FDRs in the Lab 1 clinical data.

^‡

Many of the clinical FPs were systematic errors in the OTOA and CFTR genes, which were tested in many patients.

^§

The authors did not provide a count of unique variants. For the van den Akker study, it was calculated from the data provided.

^¶

CIs were calculated based on data from the publication. No such statistics were provided by the study authors.

^‖

The lack of FPs may have been a result of aggressive filtering, which can remove clinical TPs as well as FPs.

^∗∗

Only unique 1KG variants were analyzed. The results from the updated 1KG data mentioned in this article are described.

^††

Most of the indels in patients were intronic and homopolymer associated. These are generally not clinically significant.

^‡‡

The relatively high FDRs in this data set may have been a result of under-filtering, which can also affect CIs.