Eleven SNVs (including rs12616219 near TMEM182 with P = 5.49 × 10−8, and the rare variant, rs141611945) were taken forward for replication in independent samples (Table 1). The latest release of European UK Biobank individuals not included in the discovery stage (smoking initiation, n = 275,596; smoking cessation n = 123,851; CPD n = 80,015; pack-years n = 78,897), was used for replication of the common variants (Fig. 1). Five of the common variants replicated (four for smoking initiation and one with CPD and pack-years) at P < 0.0045. Two coding variants (rs11539157, rs1190736) were predicted to be ‘probably damaging’ by PolyPhen-2 and FATHMM. The remaining five SNVs were at least nominally associated (P < 0.01) in the replication samples and had consistent direction of effect across discovery and replication. Replication for the rare variant rs141611945 could not be carried out in UK Biobank as the SNV nor its proxies (r2 > 0.3) were available. Thus we initiated replication in African American samples of the COGA (n = 476) and HRS (n = 961) cohorts (overall MAF≈0.01). The direction of
