Using a population-based linkage analysis approach to identify transcript QTL in skeletal muscle tissues in a founder population. W.-C. Hsueh, S. Kobes, R. L. Hanson PECRB, NIDDK, NIH, Phoenix, AZ.
Linkage analysis offers an unbiased way to identify QTLs and it may identify loci with multiple and/or rare variants that are not amenable to association studies. However, its power is limited by the need to calculate identical-by-descent allele sharing (%IBD) in known relatives. Recent development in statistical methods to calculate %IBD for all possible pairs of subjects using actual genotypes (e.g. GWAS SNP data) could boost the power for linkage studies in founder populations greatly by providing more precise %IBD estimates, including for pairs of subjects with no known relationship. We assessed its utility to identify eQTLs in muscle tissue. Skeletal muscle biopsies were obtained from 149 healthy Pima Indians. Transcript levels were measured on the Affymetrix Human Exon 1.0 ST Array. Both genome-wide and gene-specific %IBD for all pairs of subjects were estimated based on ~400,000 SNPs with Beagle. We defined the gene-specific region to include SNPs within a given transcript and 200kb of its coding region. The mean %IBD for unrelated pairs was 0.021, comparable to that for 2nd cousins, consistent with a founder population. We conducted linkage analyses to identify cis-eQTLs, adjusting for age, sex and the 1st principal component from our GWAS. Permutation analyses showed that a LOD score 2.07 was equal to an empirical false discovery rate <0.05. In cis-linkage analyses of 16,840 core autosomal transcripts with SNP data, we identified 188 loci with LOD 2.07 (mean: 3.742.04, range: 2.08-13.81). The mean effect size of the cis-elements at these loci was 30%10% (range: 21%-100%). To assess power, we selected a SNP at the midpoint of each transcript region and simulated a hypothetical QTL with specified effect size at each of 15,451 unique SNPs. We conducted power analyses using the %IBD estimates based on both all SNP data (n=11,026 pairs) and self-reported relationships (n=234 pairs). We have 80% power to detect QTLs with effect size 48% for linkage analysis using empirically estimated %IBD, compared to 3.4% power using pedigree-based %IBD estimates. In summary, using one of the largest samples for such studies in muscle, we identified 188 cis-eQTLs, which can be prioritized for follow-up association studies with transcripts and related clinical outcomes. Our data also suggest that population-based linkage studies in founder populations using empirically derived %IBD may provide much greater power than using pedigree-based estimates.
You may contact the first author (during and after the meeting) at