a These authors contributed equally to this work ABSTRACT 24Computational advances have fostered the development of new methods and tools to integrate 25 gene expression and functional evidence into human-genetic association analyses. Integrative 26 functional genomics analysis for altered response to alcohol in mice provided the first evidence 27 that multi-species analysis tools, such as GeneWeaver, can identify or confirm novel alcohol-28 related loci. The present study describes an integrative framework to investigate how highly-29 connected genes linked by their association to tobacco-related behaviors, contribute to individual 30 differences in tobacco consumption. Data from individuals of European ancestry in the 31 UKBiobank (N=139,043) were used to examine the relative contribution of orthologs of a set of 32 genes that are transcriptionally co-regulated by tobacco or nicotine exposure in model organism 33 experiments to human tobacco consumption. Multi-component mixed linear models using 34 genotyped and imputed single nucleotide variants indicated that: (1) variation within human 35 orthologs of these genes accounted for 2-5% of the observed heritability (meta h 2 SNP-Total =0.08 36 [95% CI: 0.07, 0.09]) of tobacco/nicotine consumption across three independent folds of 37 unrelated individuals (enrichment ranging from 0.85 -2.98), and (2) variation around (5, 10, 15, 38 25, and 50 Kb regions) the set of co-transcriptionally regulated genes accounted for 5-36% of the 39 observed SNP-heritability (enrichment ranging from 1.60 -31.45). Notably, the effects of 40 variants in co-transcriptionally regulated genes were enriched in tobacco GWAS. These findings 41 highlight the advantages of using multiple species evidence to isolate genetic factors to better 42 understand the etiological complexity of tobacco and other nicotine consumption. 43 44 45 However, the GWAS approach is not without limitations. For example, examination of genome-53 wide variation requires a stiff penalty for multiple comparisons leading to the need for 54 increasingly large sample sizes. The requirement of sample sizes in the 100's of thousands to 55 millions (i.e., mega-GWAS) exerts pressure on the depth of phenotyping that may be done (i.e., 56 more intensive and costly phenotypes are untenable for Mega-GWAS studies). Additionally, 57SNPs implicated by GWAS are not always readily associated with gene function. In fact, a 58 majority of GWAS hits fall in non-coding or intergenic regions 1 . Linkage disequilibrium allows 59 for a relatively sparse coverage of the genome to be maximally informative, but simultaneously 60 limits the immediate "translatability" of the signals (i.e., a SNP identified by GWAS may be a 61 proxy for a causative SNP some genomic distance away). In sum, while GWAS findings have 62 become increasingly reproducible as sample sizes increase, it has become increasingly evident 63 that additional sources of data (e.g., gene regulatory and epigenetic data 2 ) are needed to 64 understand how subtle SNP effects in...