Systematic augmentation of GWAS with network propagation
Recent studies have shown that a comprehensive protein interaction network is critical for network propagation efforts9. Here, we combined the International Molecular Exchange physical protein interaction dataset19 from IntAct (protein–protein interactions)20, Reactome (pathways)21 and SIGNOR (directed signaling pathways)22. To facilitate re-use of these data (referred to as ‘OTAR interactome’) we have made the data available via a Neo4j Graph Database (ftp://ftp.ebi.ac.uk/pub/databases/intact/various/ot_graphdb/current). The physical interactions were combined with functional associations from the STRING database (v.11)23 to give a final network containing 571,917 edges connecting 18,410 proteins (nodes) (Fig. 1a). GWAS trait associations were mapped to genes using the locus-to-gene (L2G) score from Open Targets Genetics, a machine learning approach that integrates features such as SNP fine-mapping, gene distance and molecular quantitative trait locus (QTL) information to identify causal genes (Fig. 1b)11. Genes with