Predicting functionally important SNP classes based on negative selection

BMC Bioinformatics

Table 5 Comparison of the distribution of derived allele frequencies (DAF) for SNPs within several classes.

class	Perlegen population	class DAF median	ancestral repeats		genome
			DAF median	p -value	DAF median	p -value
constrained elements	AFR	0.174	0.196	3.6 × 10^-5	0.205	< 1 × 10^-8
	EUR	0.188	0.208	3.4 × 10^-4	0.229	< 1 × 10^-8
	CHN	0.174	0.208	8.7 × 10^-3	0.217	< 1 × 10^-8
constrained elements 1 kb from genes	AFR	0.174	0.196	3.2 × 10^-3	0.200	< 1 × 10^-8
	EUR	0.190	0.208	5.4 × 10^-3	0.229	< 1 × 10^-8
	CHN	0.188	0.208	6.0 × 10^-2	0.217	< 1 × 10^-8
constrained elements 100 kb from genes	AFR	0.174	0.196	1.1 × 10^-2	0.200	1.0 × 10^-8
	EUR	0.205	0.208	9.6 × 10^-3	0.229	1.1 × 10^-6
	CHN	0.188	0.208	8.3 × 10^-2	0.217	8.5 × 10^-7
constrained elements outside of genes	AFR	0.174	0.196	3.0 × 10^-3	0.196	< 1 × 10^-8
	EUR	0.188	0.208	5.3 × 10^-3	0.229	< 1 × 10^-8
	CHN	0.188	0.208	6.1 × 10^-2	0.217	< 1 × 10^-8
H3K79me3	AFR	0.174	0.196	8.0 × 10^-3	0.200	< 1 × 10^-8
	EUR	0.208	0.208	1.6 × 10^-2	0.229	1.7 × 10^-7
	CHN	0.188	0.202	1.2 × 10^-1	0.217	5.0 × 10^-8

Classes presented are 1) constrained elements, 2) constrained elements at least 1 kb from the closest gene, and 3) constrained elements at least 100 kb from the closest gene, 4) constrained elements outside of genes, and 5) H3K79me3 regulatory attributes in comparison with the genome as a whole and ancestral repeats. We perform a Mann-Whitney U-test to compare the DAF distribution for SNPs in constrained elements, constrained elements outside of genes, and H3K79me3 regulatory attributes with that of the genome and that of ancestral repeats. We list the resulting p-values as well as the median DAF for each class.

ISSN: 1471-2105