For the two GWS variants that emerged in the meta-analysis of Yale-Penn cohorts, we used logistic regression via generalized estimating equation implemented in the R package GWAF15, to correlate the two GWS variants’ genotypes from COGA with Ca-HL (defined exactly as for Yale-Penn) and adjust for sex, age, the first three PCs of ancestry, and array types. As BMI was not available in COGA, it was not included as a covariate in the analysis. The association results for these two GWS variants were combined with the summary statistics of Yale-Penn 1 and 2, and COGA in the final meta-analysis using PLINK1.9.
