A blood- and brain-based EWAS of smoking

DNA methylation offers an objective method to assess the impact of smoking. In this work, we conduct a Bayesian EWAS of smoking pack years (n=17,865, [~]850k sites, Illumina EPIC array) and extend it by analysing whole genome data of smokers and non-smokers from Generation Scotland (n=46, [~]4-21 million sites via TWIST and Oxford Nanopore sequencing). We develop mCigarette, an epigenetic biomarker of smoking, and test it in three British cohorts. Results of brain- and blood-based EWAS (nbrain=14, nblood=882, >450k sites, Illumina arrays) reveal several loci with near-perfect discrimination of smoking status, but which do not overlap across tissues. Furthermore, we perform a GWAS of epigenetic smoking, identifying several smoking-related loci. Overall, we improve smoking-related biomarker accuracy and enhance the understanding of the effects of smoking by integrating DNA methylation data from multiple tissues and cohorts.