Genetic architecture of brain gene and isoform expression
The intricate molecular mechanisms occurring within the human brain are under tight genetic control. Existing work has examined genetic influences mainly at the gene-level and thus the extent to which isoform expression in the human brain is influenced by genetic factors is unexplored. This may be important as isoforms are significantly more diverse in the brain than in other tissues. Here, we investigate the genetic architecture of 24,905 gene and 93,293 isoform-level expression by integrating the genotype array and frontal cortex bulk RNA-seq data from 855 unrelated European individuals. We find the average genome-wide SNP-based heritability (h2SNP) estimates to be comparable between gene and isoform expression (0.21 and 0.18, respectively) with the average proportion of h2SNP from cis-SNPs being 17% and 12%. Multivariate variance components linear mixed model analyses reveal substantial shared genetic influences among isoforms and that genetic correlation typically mirrors phenotypic correlation. Importantly, many genes with increased transcript complexity are only heritable at the isoform-level, motivating the use of isoform-resolution analyses to fully uncover the genetic effects on gene expression.