Large-scale GWAS meta-analysis consisting of trans-ethnic samples identifies various genetic signals on BMI

Abstract

Due to the development of computational power and statistical theories, Genome-wide association studies (GWAS) have constantly been improved to gain higher power with reduced bias. GWAS identify hundreds of susceptibility loci body mass index in various populations such as European-ancestry, or Asian groups. Meta-analysis enables us to incorporate statistical results from various studies to detect more genetics signals in GWAS, as well as discover different signals from cis- or trans-ethnic groups. Here we combined data from three sources of large-scale genetics studies: UK Biobank, GIANT consortium, and a famous Japanese study. Among over two million candidate SNPs, we successfully detected 686 significant SNPs after Bonferroni correction (P < 2.5*10^-8), with most of them being detected previously. The top five SNPs are: “rs1558902” (P value = 2.394*10^-36), “rs1421085” (P value = 4.152*10^-36), “rs2237897” (P value = 2.542*10^-32), “rs2237896” (P value = 3.966*10^-32), “rs7202116” (P value = 2.702*10^-31). Although the total number of variants identified by the meta-analysis is lower than the Japanese population-based association study, meta-analysis successfully identifies several new loci not captured by the single-group association study. We also explored the original summary statistics datasets and conducted analysis to compare the statistical results from different populations separately.