Genome-wide association studies (GWAS) provide a rich source of disease-associated genomic loci. Nevertheless, it remains a long-standing challenge to link these loci to targetable causal genes. The Open Targets Genetics Portal addresses this problem by interpreting manually curated associations from the GWAS catalog, as well as independent signals from GWAS with publicly available summary statistics, most importantly the UK Biobank GWAS data (7,24,25). The Genetics Portal performs fine-mapping to narrow down the likely set of causal variants at a given trait-associated locus and to identify the potential causal gene for a particular association. The recently added locus-to-gene score (L2G) uses machine learning to prioritise causal genes by integrating fine-mapping credible sets, QTL colocalisation and functional genomics data. This method can pinpoint causal connections between loci and distant genes, and can predict multiple causal genes, a significant improvement over approaches based on gene distance to lead SNPs.
