We also performed a meta-analysis combining the discovery and replication samples (up to 622,409 individuals). LD score regression showed that the λ (intercept) for all traits was ~1.00, which indicated that confounding factors inflating the results was not an issue [47, 48]. The combined analysis identified 35 additional novel SNV-smoking trait associations, 33 with smoking initiation, one with CPD and one with smoking cessation at P < 5 ×10-8 (Table 2). We note that among our four SNVs that did not replicate, rs216195 (in SMG6) was genome-wide significant in the combined meta-analysis of discovery and replication studies (P = 2.41 × 10−9; Table 2).
