Pairwise-discordance of SNP calls between RNA-seq and the genotyping chip was assessed for all possible combinations of RNA-seq samples and DNA genotyping samples. Discordance was calculated using the variant tools software64, which reports the fraction of discordant sites out of the total number of sites where both samples report a genotype. The basic approach for calling a match was to plot the discordance values across all samples, for an all-by-all comparison, and look for a bimodal distribution with an obvious cutoff point (consistent with pairs that should match and all other pairs which do not). Indeed, all of the distributions were bimodal with regions of zero frequency in between the two peaks. The distributions of discordance values were different for RNA-RNA vs. RNA-DNA. For RNA-RNA sample matches, we called matches as instances where two samples had less than 15% discordance from each other; for DNA-RNA matches, the cutoff was 25%. We verified RNA-DNA matching within samples. Finally, we predicted gender for each sample based on the fraction of total reads aligning to the Y chromosome; if the log(fraction) was ≤
