Sixth, we observed that D and the posterior probability of “mismatch” were highly correlated, and reasoned that D might be useful for dropping additional low-quality samples. To determine the optimal threshold for D, we successively dropped individual samples according to D, and recomputed the intraclass correlation coefficient (ICC)-based estimate of heritability 2(ρ̂MZ − ρ̂DZ) and accompanying p-values 81 for all transcripts using covariate-residualized expression data. A Benjamini-Hochberg false discovery-rate q-value for transcripts was computed using p.adjust in R (v.2.14). Dropping 19 samples with the lowest D values resulted in the largest number of significant transcripts (q < 0.10) (Supplementary Note). This choice was largely robust to the q threshold in the range q=0.05 – 0.20, and to the use of unnormalized expression data.
