recently listed as candidate signals for smoking behaviour independent of those yielded by the known loci rs16969968 or rs588765 in CHRNA5 31. Further zooming into this genomic region through SNP conditional analysis and molecular experiments is desirable to completely disentangle which of these signals are independent, as well as for gene function validation 16. Third, this study is based on summary statistics obtained by the 16 TAG participating groups in GWASs, which controlled for the effects of age and sex. It is a limitation of the present study that we could not consider other covariates, which may have an effect on the phenotypes of interests or on the genetic effects on the present phenotypes. Possible covariates include cohort, the availability of tobacco, and social pressure to start/quit smoking. Finally, we emphasize that the present results are based on a sample limited to individuals of European ancestry. Whether the biological pathway and genes identified here will replicate in populations of other ancestries remains to be established.
