Integrated genome-wide analysis of human facial morphology

The human face is a highly multipartite structure resulting from the intricate coordination of multiple factors. The high heritability of facial morphology has long been appreciated, yet little is known about the contributions of specific genes. Knowledge of the genetic architecture of facial morphology is important for understanding craniofacial morphogenesis and how these processes contribute to craniofacial disorders. Studying facial genetics may also provide a basis for DNA-informed facial prediction, which has several real-world applications. Genome-wide data on well-characterized human cohorts has great potential for generating novel insights in the post-Genome-Wide Association Study (GWAS) era. Moving beyond the conventional single variant-single trait association in GWAS, this study analyzed existing genome-wide data using three different approaches to glean insights into facial morphology, by leveraging state-of-the-art advances in 3D facial phenotype modeling and multivariate statistical approaches. Specifically, analysis in Aim 1 for the first time demonstrated the contribution of rare and low-frequency coding variants in facial variation, with eight genes being significant associated, one of which (NECTIN1) had known craniofacial function. Transcriptome-wide association analysis in Aim 2 extended a previous GWAS effort by refining potential causal genes for future functional characterization. Findings from this aim also pointed to several novel candidate genes. Finally, analysis in Aim 3 explored the role of Variance Quantitative Trait Loci and highlighted the importance of studying facial variability in addition to facial mean differences in gene discovery and mechanistic exploration. The variance prioritization strategy adopted in this aim also demonstrated its advantage in detecting gene by gene interactions involved in facial morphology. These results expanded our understanding of the genetic basis of normal-range facial morphology and will have important implications for future studies. The application of several recently developed statistical tools also helped to evaluate and generalize their utility to multivariate settings and identify their limitations. This study had public health relevance. Our findings can help provide a roadmap for understanding the genetic underpinnings of craniofacial morphogenesis and birth defects, pave the way for advances in personalized prevention and therapeutics of related conditions, and inform DNA-based facial prediction for clinical and forensic application.