Gene-level heritability analysis explains the polygenic architecture of cancer
Genome-wide association studies (GWAS) have found hundreds of single nucleotide polymorphisms (SNPs) associated with increased risk of cancer. However, the amount of heritable risk explained by these variants is limited, thus leaving most of cancer heritability unexplained. Recent studies have shown that genomic regions associated with specific biological functions explain a large proportion of the heritability of many traits. Since cancer is mostly triggered by aberrant genes function, we hypothesised that SNPs located in protein-coding genes could explain a significant proportion of cancer heritability. To perform this analysis, we developed a new method, called Bayesian Gene HERitability Analysis (BAGHERA), to estimate the heritability explained by all the genotyped SNPs and by those located in protein coding genes directly from GWAS summary statistics. By applying BAGHERA to the 38 cancers reported in the UK Biobank, we identified 1, 146 genes explaining a significant amount of cancer heritability. We found these genes to be tumour suppressors directly involved in the hallmark processes controlling the transformation from normal to cancer cell; moreover, these genes also harbour somatic driver mutation for many tumours, suggesting a two-hit model underpinning tumorigenesis. Our study provides new evidence for a functional role of SNPs in cancer and identifies new targets for risk assessment and patients stratification.