We next regressed out known biological factors (local ancestry, age, sex), potential batch effects and other unknown biological factors (top five principal components of DNA methylation) for each VMR. We used PST estimates18 to provide a measure of proportion of overall gene expression variance explained by between-population differences. PST values ranged from 0 to 1, where values close to 1 imply that the majority of expression variance is due to differences between populations. We defined ΔPST as the difference between PST values before and after regressing out the effect of VMRs associated with each gene, quantifying the proportion of ancestry-associated DEGs probably due to environmental exposure. Across brain regions, we found that the average ΔPST was 15% (12.2%, 14.4% and 18.3% for the caudate nucleus, DLPFC and hippocampus, respectively, Fig. 5c). Altogether, these results imply that unknown environmental exposure reflected in DNA methylation contributes relatively little to the observed, primarily immune-related expression differences in our BA neurotypical sample.
