SNPest: a probabilistic graphical model for estimating genotypes

BMC Research Notes

Table 4 Results on low depth, diploid data from human chromosome 20

Program	#SNPs	SNP rate	dbSNP	SNPest	Excl.	Indels	dbSNP	Homo:hetero
SNPest	14,159	0.02%	99.42%	100.00%	0.13%	454	59.03%	0.64
FreeBayes	3,175	0.01%	98.90%	5.32%	2.58%	330	60.91%	1.12
SAMtools	65,120	0.11%	99.01%	99.46%	1.76%	6,918	60.18%	0.66
GATK	54,441	0.09%	99.44%	97.84%	1.17%	7,773	60.77%	1.09
GeMS	73,694	0.13%	87.59%	99.24%	17.95%	N/A	N/A	0.62

The results from SNPest (without using the reference genome), FreeBayes, SAMtools with bcftools, GATK’s HaplotypeCaller and GeMS are shown. For each method, we report the number of high quality SNPs, the SNP rate, the fraction overlap with dbSNP 139, the fraction of SNPest predictions in common, the fraction of exclusive SNPs only predicted by this method, number of insertions/deletions, fraction of insertions/deletions found in dbSNP 139, and homozygous:heterozygous ratio for SNPs.

ISSN: 1756-0500