Figure 4a shows the effect of sample size on local eQTL identification. The figure includes nearly all published blood-derived eQTL studies7,8,15,20,31,48–52 (comparisons to the large meta-analysis of Ref. 29 described separately below), the full NTR data (n=2,494), and random subsamples of our data. We reanalyzed the datasets using a common QC pipeline on inverse quantile normalized data19 (except where unavailable8,15). For comparison, we selected a set of unrelated twins (1,263 individuals) and performed local eQTL mapping on random subsets of varying sample size, using fewer covariates (i.e., no blood counts or SNPs) and ~600K genotyped SNPs. We also evaluated our robustness approaches (normal quantile transformation, and normal quantile transformation with SNP minor allele frequency, MAF > 0.005 or > 0.01 in each subsample). For local eQTLs, there was little difference among the transformations.
