Our analysis of type 2 diabetes in South Asians used European summary association statistics from the DIAGRAM data set (described above), South Asian summary statistics data from the South Asian Type 2 Diabetes (SAT2D) Consortium (Kooner et al., 2011), and South Asian genotypes and phenotypes from UK Biobank (see Web Resources) as test data (row 5 of Table 1). The SAT2D data set consists of 5,561 South Asian type 2 diabetes cases and 14,458 South Asian controls for which we summary statistics for 2,646,472 imputed SNPs were available. The UK Biobank test data consists of 1,756 unrelated samples of South Asian ancestry (272 type 2 diabetes cases and 1,484 controls), genotyped at 608,878 SNPs after QC, with the following self-reported ethnicity distribution: 52 Bangladeshi, 1,301 Indian and 403 Pakistani. We removed one individual from each pair of relatives with relatedness greater than 20% (n=30). We performed a PCA analysis using EIGENSTRAT (Price et al., 2006) (see Web Resources) to identify and remove genetic outliers, but did not identify any outliers. We analyzed 208,400 SNPs present in the DIAGRAM, SAT2D and UK Biobank data sets after removing A/T and C/G SNPs to eliminate potential strand ambiguity.
