Introduction
Obesity is heritable, predisposes to many diseases and is commonly defined using body mass index (BMI). A large number of common and low frequency variants have been identified to be associated with BMI using GWAS. However, these common variants only explain a small fraction of heritability and a vast majority of variants in the human genome are rare. Because GWAS did not well captured rare variants (RVs), rare variant association analysis have so far not been very successful for detecting BMI-associated RVs. Ongoing large-scale whole genome sequencing (WGS) studies, such as the multi-ethnic NHLBI Trans-Omics Precision Medicine (TOPMed) Initiative, enable assessment of associations between BMI and rare variants (RVs) across the genome.
Hypothesis
Rare variant (RV) aggregations are associated with BMI.
Methods
We performed analysis of TOPMed Freeze 5 data from deep whole genome sequencing (>30X coverage) of 45,159 individuals across 20 TOPMed studies. After QC, a total of 326M variants were identified, 314M of which were rare variants (MAF<1%). For rare variant set association analysis, we applied our newly developed variant-Set Test for Association using Annotation infoRmation (STAAR) method, which dynamically incorporates multi-faceted variant function annotations to perform both whole genome gene-centric functional category based analysis and genetic region analysis using sliding windows across the genome.
Results
For RV-set gene-centric analysis, STAAR identified 2 genome-wide significant associations (P<0.05/20000= 2.5×10^(-6)) with BMI, including putative Loss-of-Function RVs in GNA14 and missense RVs in CREBRF. After conditioning on known BMI-associated variants, both aggregation tests remained significant. In sliding window analysis, one region in the gene FANCM achieves genome-wide significance defined using Bonferroni correction for the number of tests (P<0.05/5355476=9.3×10^(-9)). This association remains significant after adjusting for known BMI-associated variants.
Summary
Several novel RV-sets associated with BMI were identified using the TOPMed WGS Freeze 5 data. By incorporating variant functional annotation, STAAR empowers RV association analysis in WGS.