Polygenic architecture of human body size and proportion

Abstract

Over the last 15 years, genome-wide association studies (GWAS) have discovered thousands of genetic associations with human phenotypes. However, as most of these associations lie in non-coding regions and are typically spread over a group of highly correlated variants, it remains difficult to draw clear connections from the variant through relevant biological mechanisms to the associated phenotype. The field of genetics has endeavored to address this deficiency in many ways. I have built upon prior findings using various approaches, integrating genetic data from multiple phenotypes and ancestries, to bridge the gap between association and genetic architecture. My work focuses on phenotypes related to skeletal growth and proportion. I first explored the genetic basis of multiple growth-related proteins. Here, I identified two protein quantitative trait loci (pQTL) associated with serum levels, measured in a childhood Cincinnati cohort, for IGFBP-3, IGF-2, and IGFBP-5. To better understand their effects, we explored each association’s overlap with adult height as well as related phenotypes including sitting height ratio (SHR), a measure of skeletal proportion, and birth weight (BW). Mendelian Randomization (MR) supports a causal relationship between protein levels and SHR (for an association near IGFBP3) and BW (for an association near IGFBP5) but not for height. This result suggests that the mechanism by which these proteins affect height must be through some process, perhaps local to the growth plate, not reflected in measured serum levels of these proteins. I then investigated the genetic basis of SHR, using genetic data from two ancestries to perform the largest GWAS of SHR to date. After identifying 565 independent associations (an increase from 6 in the prior publication), I observed substantial overlap between phenotypes at the level of both the associated loci and implicated genes and pathways. Using fine-mapping, I classified height associations by their effect on body proportion, and showed that those fine-mapped credible sets affecting both height and body proportion are enriched for critical genes for growth. Additionally, these fine-mapping results

enabled me to identify instances where effects on height and body proportion differed across different ancestries. Lastly, I used various approaches to understand the genetic structure underlying height and other anthropometric traits. I first quantified the extent to which biological pathways implicated by gene set enrichment analysis (GSEA) were “saturated” across increasingly large height GWAS. I then identified genes and gene-sets enriched among height GWAS results, and performed similar analyses in collaboration with GIANT working groups focused on body mass index and waist-hip ratio, and developed comparative GSEA, an approach to identify enriched gene sets that differ between input GWAS. In addition, I quantified levels of population stratification present in height GWAS samples, and re-examined evidence of natural selection acting on loci identified in height GWAS. Together, the findings and methods described in this dissertation expand our understanding of the biology and genetic architecture underlying measures of human body size and proportion, and contribute novel methodological approaches to understanding correlated phenotypes.