For inclusion in the Catalog, studies and associations must meet strict criteria (www.ebi.ac.uk/gwas/docs/methods); studies must include an array-based GWAS and analysis of >100 000 SNPs with genome-wide coverage, while SNP-trait associations must have a P-value <1 × 10−5. Studies identified through an automated PubMed literature search are reviewed to select those matching the inclusion criteria. In order to summarize the GWAS literature in a consistent manner in the Catalog, each GWAS analysis is entered separately into the curation system using standards for describing ancestry and traits, and following extensive extraction guidelines at the study and SNP levels. Trained curators, with expertise in deciphering study design, read each paper to determine how to represent the GWAS analyses in the most scientifically accurate and accessible manner. The curators assess whether any appropriate replication analyses were performed; the number and ancestry of samples included in the analysis; and the traits analyzed. All traits are mapped to terms from the Experimental Factor Ontology (EFO; 17). This enables structured querying and visualization of data in the GWAS diagram (Figure 1) e.g. searching for ‘cardiovascular disease’ displays all associations both with this specific trait and its sub-traits, including for example ‘myocardial infarction’ and ‘coronary artery disease’.
