Functional Annotation of GWAS Loci Using Public Transcriptome and Epigenome Datasets Reveals Non-Coding Genes and Regulatory Elements Which May Contribute to BMI.

Genome-wide association studies have identified numerous genetic variants statistically significantly associated with body mass index (BMI). However, the functional mechanisms underlying most associations between single nucleotide polymorphisms (SNPs) in non-coding regions and BMI remain poorly understood. Here, we implemented an integrative 7-criterion quantitative scoring system (gene localization, histone modifications, transcription factor binding sites (TFBS), SNP clouds, tissue expression patterns, evolutionary conservation, and COVID-19 associations) to prioritize putative functional loci among 94 BMI-associated SNPs. Six SNPs resided within long non-coding RNA (lncRNA) genes: rs2245368 (exonic, DTX2P1-UPK3BP1-PMS2P11), rs2033529 (exonic, LINC00951), rs2836754 (intronic, ETS2-AS1), rs2815752 (intronic, LINC02796), rs17203016 (intronic, MYOSLID-AS1), and rs7239883 (intronic, LINC00907). We prioritized them because they are located within lncRNA gene bodies and therefore showed stronger functional support compared to other variants which only had non-coding regulatory elements in their vicinity. Notably, rs1928295 exhibited strong GATA2 binding evidence, while rs13201877 had extensive transcription factor occupancy (64 factors). Multiple variants demonstrated putative regulatory potential through epigenomic evidence, including DNase I hypersensitivity and cell-type-specific chromatin accessibility. We show that most BMI risk alleles are not human-specific and are conserved across primates. Our findings suggest putative candidate non-coding regulatory elements in BMI and provide prioritized obesity-associated non-coding variants for functional validations.