Abstract
We present a novel approach to detect potential cis-acting regulatory loci that combines the functional potential, an empirical DNase-seq based estimate of the allele-specificity of DNase-I hypersensitivity sites, with kernel-based variance component association analyses against expression phenotypes. To test our method we used public ENCODE whole genome DNase-I sequencing data, from a single sample, to estimate the functional potentials of the subset of 10,552 noncoding heterozygous single-nucleotide polymorphisms (SNPs) that were also present in the Genetic Analysis Workshop 19 (GAW19) family-based data set. We then built two covariance kernels, one nonweighted and one weighted by the functional potentials, and conducted kernel-based variance component association analyses against the 20,527 transcript expression phenotypes in the GAW19 family-based data set. We found signals of potential cis-regulatory effects, that surpassed the Bonferroni significance threshold, for ten transcripts. Stepwise removal of the cis-located SNPs from the weighted kernel lead to the disappearance of the association signal from our top transcript hit. We found compelling evidence of allele-specific cis-regulation for four transcripts using both kernels, and our results agree with previous research that suggests the involvement of specific cis-located variants in the regulation of their neighboring gene.
Highlights
Variation found in noncoding regions of the genome is much more abundant and, perhaps, even more relevant than coding variation for certain human traits, but its biological meaning is hard to assess [1]
Data set We used single-nucleotide polymorphism (SNP) dosages from 959 genotyped individuals, transcript expression levels from 647 of those individuals, and the genealogies (1389 individuals in 20 families) that were provided as part of the Genetic Analysis Workshop 19 (GAW19) family-based data set [10]
We were able to measure the allele-specific chromatin accessibility and estimate the functional potential (FP) for 48,236 (1.99 %) of those heterozygous SNPs but only 10,618 (22 %) of them were present in the GAW19 dosages
Summary
Variation found in noncoding regions of the genome is much more abundant and, perhaps, even more relevant than coding variation for certain human traits, but its biological meaning is hard to assess [1]. It has been noticed that between 34 and 88 % of the disease-associated variants detected by genome-wide association studies (GWAS) appear to cluster in noncoding regions of the genome, in DNase-I hypersensitivity sites (DHSs) [2], and that some of the DHSs exhibit allelespecificity [2,3,4]. We have been investigating a systematic approach that uses DHSs to determine if noncoding single-nucleotide variation changes the local allele-specific chromatin accessibility, something that would indicate a potential regulatory role for a variant [6]. We have developed a variance component based burden test to determine the contribution of localized relationship kernels to the trait variance [7, 8]. We test if by combining both lines of research we could detect potential cis-acting regulatory loci.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have