This interface presents results from a meta-analysis of type 2 diabetes genetic associations from exome sequencing data for 86,254 samples (29,434 T2D cases and 56,820 controls). The analysis was performed by Laura Raffield, Yin Xianyong, Andrew Ruttenberg, Peter Dornbos, Ryan Koesterer, and Jason Flannick.
The focus was on rare coding variants, which were analyzed using burden testing via previously described variant grouping methods. Briefly, variants were grouped into seven differing masks based on the likelihood of predicted deleteriousness to protein structure and function, ranging from a mask with only high-confidence loss-of-function variants to a mask that includes all rare missense variants. Following, the p-values were combined and corrected across the seven differing test using the minimum p -value method. For further details, see Flannick et al., Nature, 2019 (PMID: 31118516).
Please note that, due to data quality issues, the gene-level results are currently not complete and should be analyzed with caution. We believe these results to be accurate, but they produce weaker associations for known positive controls. We are currently investigating possible reasons as to why this could be the case, including phenotypic or genetic heterogeneity between cohorts. See page 4 in the whitepaper for a full discussion of these issues.
Download the whitepaper with full details of the analysis, including single-variant association analysis for T2D and 24 quantitative traits.