Comparison of genotype clustering tools with rare variants

BMC Bioinformatics

Table 3 Call concordance with the 1000 Genomes Project (heterozygous and homozygous rare allele)

Tool	NA12763 _R		NA12763 _R1		NA12763 _R2
	Rate	Number	Rate	Number	Rate	Number
GenCall (original)	0.990155	12,087	0.990650	12,085	0.988351	12,104
GenCall (optimized)	0.994848	12,035	0.994931	12,035	0.995014	12,033
GenoSNP (original)	0.767847	16,179	0.514147	18,590	0.137233	19,507
GenoSNP (optimized)	0.967032	12,042	0.970397	11,992	0.963052	12,071
optiCall	0.962401	12,527	0.962478	12,526	0.958045	12,585
zCall	0.994053	12,107	0.994136	12,107	0.993970	12,107

Call concordance and number of compared markers for the three control replicates when compared to the 1000 Genomes Project. For each dataset (i.e. tool), only genotypes called as heterozygous or homozygous of the rare allele (according to the allele frequency computed using the corresponding dataset) were kept for analysis. The following six tools were compared: GenCall (original and optimized cluster files), GenoSNP (original and optimized), optiCall (without excluding markers failing Hardy-Weinberg) and zCall.

ISSN: 1471-2105