A Meta-Analysis of Genome-Wide Association Results in Type 2 Diabetes

This page describes a meta-analysis of association results publicly available by the Diabetes Genetics Initiative (DGI) and the Wellcome Trust Case Control Consortium (WTCCC), primarily meant to illustrate some of the technical details of performing a genome-wide meta-analysis.

We emphasize that a much more extensive 3-way meta-analysis (including the FUSION genome scan) through HapMap-based imputation was described by Zeggini et al. (2008).

Note: you may have to contact the authors of these data sets directly to obtain the data.

In this meta-analysis we were able to directly combine association results across the DGI and WTCCC studies -- without the need for imputation, since both studies used the Affymetrix 500K genotyping platform. Starting from the original association results, we eliminated assocation results for multimarker tests (SNPs starting with "i_") in DGI, and kept only SNPs present in both data sets.

In the WTCCC data, we flipped all SNPs that were annotated by Affymetrix as being on the (-) strand of build 35 (click here for a list of these SNPs). Subsequently we made sure that the listed "allele1" (in WTCCC) corresponded to "A1" (in DGI) by flipping the minor/major alleles in WTCCC when they did not match. (In the DGI data, all alleles are given relative to the (+) strand of build 35.)

For 365,918 SNPs, we converted the listed p-values ("Z_PVAL" in DGI and "frequentist_add" in WTCCC) into standard normal z-scores, and computed a joint z-score, observing the direction of effect in both studies. We weighted the contribution of corresponding z-scores of each study by their estimated effective sample size (as if it had been a symmetric study with equal number of cases and controls) in order to deal with study design differences between DGI (1464 cases and 1467 controls, including 326 discordant sibships; effective sample size = 2,522) and WTCCC (1924 cases and 2938 controls; effective sample size = 4,706). We found that 102 SNPs were monomorphic in either cases or controls (WTCCC), but kept them to allow comparison.

  • Download R script that was used to compute the meta-analytic z-scores and p-values
  • Download input file for the R script (in case you want to reproduce the calculations)

To make sure the meta-analysis was performed correctly, we computed the genomic inflation factor lambda by computing the ratio between the median of the squared z-score distribution and the expected median value of 0.455 (for a 1-df test). Lambda was 1.011 for DGI, 1.082 for WTCCC, 1.081 for the meta-analysis. Therefore, there was no evidence of additional inflation of the test statistic in the meta-analysis, although one might prefer to correct for the apparent inflation by dividing the chi-square statistics by the observed lambda in the two respective studies before performing the meta-analysis.

Questions?

You can contact the author here.

References

Diabetes Genetics Initiative of Broad Institute of Harvard and MIT, Lund University, and Novartis Institutes of BioMedical Research (2007) Genome-wide association analysis identifies loci for type 2 diabetes and triglyceride levels. Science. 316: 1331-1336.

Wellcome Trust Case Control Consortium (2007) Genome-wide association study of 14,000 cases of seven common diseases and 3,000 shared controls. Nature. 447: 661-678.

E. Zeggini, M.N. Weedon, C.M. Lindgren, T.M. Frayling, et al. (2007) Replication of genome-wide association signals in UK samples reveals risk loci for type 2 diabetes. Science. 316: 1336-1341.

E. Zeggini, L.J. Scott, R. Saxena, B.F. Voight, et al. (2008) Meta-analysis of genome-wide association data and large-scale replication identifies additional susceptibility loci for type 2 diabetes. Nature Genetics. 40: 638-645.

Credit

This study makes use of data generated by the Wellcome Trust Case Control Consortium. A full list of the investigators who contributed to the generation of the data is available from www.wtccc.org.uk.