A 26-hour system of highly sensitive whole genome sequencing for emergency management of genetic diseases

Table 2 Comparison of the analytic performance of a conventional alignment and variant calling pipeline (GSNAP with GATK minus VQSR), with a novel, extremely rapid method (DRAGEN)

Sample	SBS₁₈ yield (GB)	Site	Pipeline	Reads aligned	Alignments with mapping quality >20	Variants called	Mismatch rate	Indel rate	% Paired Reads	Strand balance	% Chimeric Reads	Rare, potentially pathogenic variants	Analytic sensitivity (GeT-RM or SNP array)	Analytic specificity (GeT-RM or SNP array)	Analytic sensitivity (full GIAB)	Analytic specificity (full GIAB)
NA12878	133	Essex	DRAGEN	99.4 %	95.48 %	4,782,970	0.0029	0.00017	99.55 %	0.500	0.69 %	658	99.93 %	99.87 %	99.69 %	99.99 %
NA12878	133	Essex	GSNAP/GATK-1.6/noVQSR	98.5 %	96.33 %	5,343,988	0.0056	0.00017	98.55 %	0.496	0.82 %	783	99.54 %	98.57 %	98.21 %	99.99 %
NA12878	65^a	CMH	DRAGEN	97.7 %	91.31 %	4,633,357	0.0060	0.00023	99.18 %	0.501	1.89 %	775	99.42 %	99.46 %	98.63 %	99.99 %
NA12878	65^a	CMH	GSNAP/GATK-3.2/noVQSR	96.2 %	92.86 %	4,571,157	0.0079	0.00021	97.55 %	0.499	1.75 %	593	97.29 %	95.35 %	95.74 %	99.99 %
UDT_173	106	Essex	DRAGEN	99.5 %	94.92 %	4,742,150	0.0034	0.00020	99.80 %	0.500	1.12 %	620	96.13 %	97.74 %	n.a.	n.a.
UDT_173	106	Essex	GSNAP/GATK-1.6/noVQSR	99.3 %	96.88 %	4,294,504	0.0034	0.00019	99.34 %	0.500	0.90 %	512	88.54 %	98.06 %	n.a.	n.a.

All runs were 18-h WGS. The NA12878 reference genotypes were NIST High Confidence calls from GeT-RM/NA12878.NIST-GIAB_v.2.18 (labeled ‘GeT-RM’) or the full GIAB dataset (labeled ‘full GIAB’). UDT_173 were results of hybridization to the Omni4 SNP array. GSNAP was version 2012.07.12, with default parameters, and GATK was version 1.6.13 or 3.2, without VQSR. DRAGEN was version 1.2. % paired, percentage of reads whose mate was also aligned; Strand balance, reads aligned to the forward strand divided by total reads aligned; % chimeric, percentage of chimeric alignments (mates >100 kb apart or on different chromosomes). ^aSingle flowcell

ISSN: 1756-994X