The discovery GWAS included genotyped as well as imputed SNPs for better coverage. Sample genotypes were phased and missing markers were imputed using the software IMPUTE2 (ref. 54 (link)). Phased data of 1,094 samples from the 1000 Genomes Project (www.1000genomes.org) were used as the reference data set. These samples are from African, African American, Asian, Caucasian and Native American populations. We excluded SNPs with an information coefficient value <0.5 and minor allele frequency <1%. For markers in the MEC data, we conducted additional stringent quality-control analyses to exclude SNPs with different minor allele frequencies in controls typed on or imputed from different genotyping platforms. SNPs that exhibited differences in minor allele frequencies between the controls genotyped with the Illumina Infinitum 660 W-Quad or Omni 2.5 M platforms at P<0.01, and/or had ORs >10 or <0.1 were excluded. The final GWAS analysis included 7,229,558 SNPs.
Free full text: Click here