After genotyping QC and imputation, 8.3 million SNPs were available for eQTL mapping in 2,494 individual twins (90.4% of the expression dataset). We evaluated multiple “predictors” of heritability, including association r2 based on the most-significant local SNP within ±1Mb with, r2 for the top distant SNP, local SNP-heritability estimation based on genetic relatedness among unrelated subjects using GCTA,46 and variance-components results from complete local identity-by-descent inference among the DZ pairs (local IBD). We computed ratios of each component to the overall h2 estimate (Supplementary Figure 3). Means and medians for r2local SNP/h2 (0.04, 0.09) were similar to those reported in the MuTHER study 8, while the ratio h2local IBD/h2 was higher (median=0.11, mean=0.30), consistent with higher explained variation when the total local contribution was considered. However, in published studies, estimates have been complicated by bias and variability in h2 estimation. MuTHER reported mean h2 in expressed genes of 0.16 (skin), 0.21 (LCLs), and 0.26 (adipose), with >20% of expressed genes displaying h2 > 0.3.8 Our study, although much larger, produced lower values of 0.14 and 12.3%. Our average h2 should
