Project Summary (Unchanged)
Genetic variations, environmental exposures, and their interactions underlie the etiology of all human diseases.
While genome-wide association studies have revealed many trait-associated genetic variants and
epidemiological studies have pinpointed myriad disease-associated environmental factors, the role of their
interactions is much less explored, mainly due to the lack of very large population cohorts, high-quality
environmental measures, and efficient tools. This proposal aims to characterize gene-environment interactions
(GEI) in both human evolution and complex traits. Genetic and polygenic adaptations to local environments
during human evolution have shaped the gene-environment relationship and the genetic architecture of complex
traits. Leveraging the growing number of ancient DNA, we will first develop and apply statistical tests to identify
genetic and polygenic responses to the Agricultural Revolution. The findings of adaptive genetic variants and
polygenic traits will inform our understanding and study of the current epidemics of complex diseases, which are
likely results of present-day gene-lifestyle mismatches. Second, to directly identify and quantify GEI in complex
traits, we will develop an efficient computational pipeline and perform large-scale interaction analysis across the
genome, phenome, and selected high-quality environmental factors in UK Biobank. All summary statistics will
be released publicly as a database on a dedicated website to fuel further explorations, such as meta-analysis
and testing for replicability across cohorts and ancestries. Third, to assist and guide future GEI studies, we will
develop the first bioinformatics tool for phenome-wide interaction study (PheWIS) of target genetic variants and
environmental exposures, enabling efficient and unbiased search for environment-modifiable phenotypic effects.
Moreover, to alleviate the multiple testing burden in GEI studies with a large number of exposures and clinical
outcomes, we will examine if Mendelian randomization analysis coupled with phenome-wide association study
(PheWAS-MR) could be an effective way to prioritize potentially causal exposure-outcome relationships, which
may increase the statistical power of detecting GEI and assist the downstream search for functional mechanisms.
Lastly, as an effort to improve the portability of polygenic score (PGS) across ancestries and subgroups with the
same ancestry, we will use simulated and empirical data to test if explicit statistical modeling of GEI could mitigate
the problem. Concurrently, we will examine if PGS-environment interaction analysis is an effective approach to
identify actionable environmental exposures that attenuate genetic risks. Overall, this proposed research will
generate new methods, computational tools, database resources, and novel insights into the general patterns of
GEI in human complex traits.