paperKB
coga / coga-kb
Help
Sign in

Chunk #19 — Results — TWAS identifies novel expression-trait associations

Source
Integrative approaches for large-scale transcriptome-wide association studies.
Embedded
yes

Text

Next, we evaluated the contribution to heritability of all expression-trait associations, including those that were not genome-wide significant (Methods,29,30). We estimated the variance in trait explained by all METSIM+YFS imputed genes ( hGE2) to be 3.4% averaged over six traits (Supplementary Table 10). We assumed independence between the two cohorts, and did not include the NTR genes because of its strong correlation with YFS. Height had the most variance attributable to the heritable genes at hGE2=7.1%. These combined estimates were consistently higher than a corresponding analysis using predictions from permuted expression (Supplementary Table 10). For the four traits with individual-level genotype and phenotype data in the METSIM (BMI, TG, WHR, INS), we estimated hGE2 directly using variance-components over the imputed expression values (Methods). On average, all significantly heritable genes in adipose + blood explained 4–6% of the trait variance (16–19% of the total trait hg2), and were largely orthogonal between the two predictions (Supplementary Table 11). The imputed expression consistently explained more trait variance than the best cis-eQTL in each gene and did not strongly depend on the cis-window size (Supplementary Table 12).