COPDGene EA with WGS data and smoking controls in COGEND EA with array data, due to high R2 difference between the WGS and array data. Thus, GAWMerge may lose some sensitivity while controlling type-I errors. There is also the potential for reduced power to detect COPD associated genetic variants here due to the missingness of lung function phenotypes in COGEND public controls, with power being reduced relative to the amount of COPD status misclassification among these controls. Third, when GAWMerge has been tested as an application of GWAS, it is limited by the MAF and genomic coverage on array genotyping technologies. Since GAWMerge extracts only SNPs within the array technology, the complete coverage of WGS (over 410 million variants within TOPMed WGS data26) is not fully utilized. Therefore, those rare variants and large insertions/deletions only detected in WGS data were lost during the extraction and merging processes (Supplementary Table 3). However, coming from a case-only dataset with array-based genotyping, the dominant scenario for use of GAWMerge, the WGS is a substantial strength, accounting for all the array genotyped variants except for technology based regional loss of variants. With our strategy of WGS data as public controls for GWAS, there will
