Association results from all six cohorts were combined using the inverse variance weighted method implemented in METAL (http://www.sph.umich.edu/csg/abecasis/metal/). In order to account for modest relatedness between the cohorts 1–4, BMI Z-scores of all six cohorts were pooled and analyzed together, using variance component method for comparison with the meta-analysis method. To evaluate the potential confounding effect of disease, 2111 community non-diabetic subjects from cohorts 1, 3 and 6, 1516 T2DM–ESRD subjects from cohorts 2 and 4, and 1362 T2DM subjects from cohorts 4, 5 and 6, were separately analyzed using linear regression and variance component methods, as appropriate. The association results were thencombined using METAL to assess the overall SNP association in non-diabetics, T2DM–ESRD and T2DM separately. For loci showing multiple associations, conditional SNP analyses that include multiple SNPs as independent variables were performed, using variance component method in the pooled samples to evaluate the independence of the association signals.
