Methylation QTL (meQTL) analyses

Genome-wide meQTL analyses were performed testing for the association between common genetic variants and DNA methylation at CpG sites in the two adipose tissue samples. We only considered SNPs that were significantly associated with DNA methylation in cis to be meQTLs. If multiple SNPs were identified for a single CpG site, we reported only the most significant SNP per CpG site (P = 5 × 10⁻⁵, as described in Grundberg et al. [72]). In total, methylation levels of 102,461 CpG sites were associated with genetic factors in cis, and 25,531 sites in trans.

We tested for adipose tissue meQTLs first by fitting a LME model regressed all the identified covariates, then performed a linear regression of the residuals on the SNPs using the MatrixeQTL R package [89]. Results from meQTL analyses are presented at a P value of 10⁻⁵ for the smoking-DMS, the smoking-DES, and at the smoking GWAS genetic variants. For meQTL analyses replicating the results from Loukola et al. [43], we applied a different threshold. Loukola et al. [43] conducted a genome-wide association study of nicotine metabolite ratio, identifying many strongly associated SNPs in a 4.2-Mb region on chromosome 19q13. Among the 158 CpG sites within that region, 16 CpG sites showed statistically significant association with 173 SNPs. We compared our meQTL findings to those from Loukola et al. [43] at a modified Bonferroni significance threshold of 1.81 × 10⁻⁵ (= 0.05/(16 × 173)) and identified SNPs that influence methylation levels at 5 CpG sites (in CYP2A7, ENGL2, and LTBP4 genes) (Additional file 2: Table S5).