Chunk #47 — Gene network inference within individuals using cross-tissue expression variation — Impact of individual gene-disrupting variants on splicing and expression
The GTEx multitissue expression data provide an opportunity to assess the real impact of protein-truncating variants (PTVs) on the human transcriptome [see also (31)]. We combined exome sequencing and RNA-seq data from 173 GTEx individuals to assess the global properties of predicted high-confidence PTVs (14). PTVs are enriched in alternatively spliced exons, with just 38.4% of high-confidence PTVs having annotation support across all reported transcripts (Fig. 7A), and only 51 to 55% supported by the major transcript of at least one tissue (for all tissues with at least 10 samples). These numbers highlight the need for careful transcript-specific assessment of functional annotation for all classes of variation. Furthermore, if we require that a fixed percentage of the dominant isoforms across all sequenced tissues support this annotation, we find that the percentage of predicted PTVs with annotation support of PTV decreases as we increase the threshold for the proportion of tissues with major transcript isoform support for PTV prediction (Fig. 7B). This highlights the need for empirically derived reference transcript sets across a broad array of tissues to enhance clinical interpretation for personal genome sequencing and disease studies. An example with clinical ramifications is shown in fig. S32.
