High-throughput detection of mutations responsible for childhood hearing loss using resequencing microarrays

BMC Biotechnology

Table 1 Overall array performance with and without application of sPROFILER to GDAS/GSEQ base calls.

	Harvard arrays with GDAS*	Harvard arrays with GDAS/sPROFILER*	Cincinnati arrays with GSEQ	Cincinnati arrays with GSEQ/sPROFILER
Number of arrays	26	26	13	13
Bases per array	25187	25187	26292	26292
Array call rate ^A	96.9%	99.6%	97.9%	99.6%
Call accuracy ^B	99.82%	99.84%	99.83%	99.88%
Total false positive rate ^C	0.18% (41)	0.15% (38)	0.16% (42)	0.11% (30)
Total false negative rate ^D	0.0016% (0.4)	0.0031% (0.9)	0.0009% (0.2)	0.0020% (0.6)
Variant false positive rate ^E	72.6% (41/57)	71.6% (38/51)	77.7% (42/54)	69.7% (30/43)
Variant false negative rate ^F	2.4% (0.4/16)	4.5% (0.9/16)	1.3% (0.2/15)	3.0% (0.6/15)
No-calls ^G	781	101	563	103
No. of exons to be sequenced ^H /Total no. of exons on array	153/196	52/196	150/180	68/180

Percentages are obtained by averaging individual percent values over all arrays.
A: Bases called/total bases on array
B: Correct calls/total calls
C: Wild-type bases incorrectly identified as variants/total calls * 100% (average raw # per array)
D: True variants incorrectly called wild-type/total calls * 100% (average raw # per array)
E: Wild-type bases incorrectly identified as variants/total variant calls * 100%
F: True variants incorrectly called wild-type/total true variants * 100%
G: Average number of bases not called per array
H: Number of exons that need follow-up sequencing to interrogate no-calls or variant calls
*: 14 Harvard arrays with full dideoxy sequencing results were used for determination of false negatives and overall accuracy. However, no-calls and variant calls across all 26 Harvard arrays were used for call rates and false positive rates.

ISSN: 1472-6750