NEW YORK (GenomeWeb) – A team from the UK has assembled a publicly accessible database called GeneATLAS, representing genetic associations gleaned from available UK Biobank data.
Researchers from the University of Edinburgh did genome-wide association studies on up to 452,000 UK Biobank participants, searching for contributors to 778 binary or non-binary traits. As they reported online today in Nature Genetics, the resulting GeneATLAS collection makes it possible to search for potential associations involving millions of directly genotyped or imputed SNPs, while providing accompanying GWAS summary statistic data.
"Unlike other currently available databases … our database includes significant and non-significant associations, thus providing an unbiased view of phenotype-genotype associations across a large number of traits within a single cohort," corresponding and co-senior author Albert Tenesa, a researcher with the University of Edinburgh's Roslin Institute and the MRC Institute of Genetics and Molecular Medicine, and his colleagues wrote.
The collection is expected to offer clues to complex trait genetic architecture, the team explained, noting that the GeneATLAS currently includes variants involved in more than 182,200 genotype-phenotype associations, as well as heritability estimates and correlations with other genetic or environmental features.
Starting with data from some 490,000 array-genotyped UK Biobank participants, the researchers used GWAS analyses to search for variants coinciding with 660 binary and 118 non-binary, continuous traits. Following quality control steps, they were left with information for 452,264 individuals.
When it came to the binary traits such as skin cancer or celiac disease, for example, they were able to assess data for almost 6,600 cases per trait, on average, though more cases were available for the hypertension analysis. Across the binary traits considered, the investigators unearthed more than 393,000 genetic associations, with many falling in and around the human leukocyte antigen immune locus.
In the case of the non-binary traits, on the other hand, the team tracked down genome-wide significant associations involving nearly 10 percent of variants in the analyses, though far more variants showed less significant ties to these continuous traits.
The researchers noted that a variant near the FTO gene locus showed the strongest average non-binary trait associations, while height represented the non-binary trait with the largest set of significantly linked variants.
With the data at hand, the team also explored everything from the heritability of specific traits to the feasibility of predicting height or other phenotypes based on SNP marker sets.
"We hope this database will be useful to those working on complex trait genetics, but also to those that do not have the expertise or capabilities to perform analyses at this scale," the authors wrote.