To evaluate the performance of GAWMerge, we used three smoking-related datasets: Collaborative Genetic Study of Nicotine Dependence (COGEND)20,21, Genetic Epidemiology of COPD (COPDGene) study22, and Evaluation of COPD Longitudinally to Identify Predictive Surrogate End-points (ECLIPSE)23. As indicated in Table 1, the three datasets have different array platforms, providing the opportunity to assess the performance of the protocol in different settings. In both COPDGene and ECLIPSE, the COPD diagnosis followed the Global Initiative for Chronic Obstructive Lung Disease severity classifications, and COPD cases were defined as severity Grade 2–4 COPD (moderate, severe, and very severe COPD)24. The study design to evaluate GAWMerge across (a) genotyping technology (ensuring no technology driven false positives), (b) type-I error (ensuring minimal false positive associations), and (c) recovery of known GWAS hits (demonstrating capture of true positives) is presented in Fig. 2.Table 1Dataset characteristics.COGENDCOPDGeneECLIPSEArray typeIllumina HumanOmni2.5Illumina HumanOmni1-Quad_v1-0_BIllumina HumanHap550v3.0Array-genotyped dataN, SNPs2,443,1791,051,295561,466Participants, total N2,6739,9622,159Ancestry group, N (%)European1,961 (73%)6,664 (67%)2,159 (100%)African American712 (27%)3,298 (33%)NASex, N (%)Males1,019 (38%)5,333 (54%)1,367 (63%)Females1,654 (62%)4,629 (46%)792 (37%)COPD diagnosis, N (%)YesNA4,280 (43%)1,764 (82%)No3,632 (36%)395 (14%)Age (mean ± SD)36.6 ± 5.659.6 ± 9.062.2 ± 8.2WGS-genotyped dataaParticipants,
