ABSTRACT
Unhealthful diet is a leading risk factor for diabetes and mortality worldwide. As the diabetes and obesity
epidemics continue to rise, so does the contribution of dietary risk factors to global disease burden. There is an
urgent need to identify which aspects of diet causally influence metabolic disease to guide more effective dietary
recommendations. Teasing apart correlation from causation remains a challenge, and while numerous
epidemiological studies have observationally linked diet to diabetes, there has been limited success with
translation to intervention studies. Normal human genetic variation has both direct and indirect effects on dietary
intake, with recent work establishing significant heritability and hundreds of genetic associations with numerous
different foods and dietary patterns. However, combining dietary traits across studies for genetic analysis
remains a challenge due to study differences in design, cultures, and preferences. We hypothesize shared
genetic influences on dietary intake can act as the common reference to identify comparable diets across studies.
In each of several cohorts, with both genetic and diet data, we will initiate new collaborations, derive quantitative
food traits and dietary patterns, and conduct genome-wide association studies (GWAS) to create homologous
GWAS datasets with study-specific dietary phenotypes and a common set of genetic markers. A series of genetic
correlation analyses will be conducted to identify comparable foods and dietary patterns across diverse studies.
Once identified, GWAS meta-analysis of comparable dietary phenotypes will improve power to detect novel and
multi-ethnic genetic associations. To elucidate the direct and indirect genetic mechanisms of dietary intake at
the locus and genome-wide levels we will conduct fine-mapping and gene prioritization, enrichment and pathway
analysis, and genetic correlation and phenome-wide association studies (PheWAS). To address limitations with
observational studies, Mendelian randomization (MR) causal inference will be performed using genetically
predicted dietary intake and publicly available GWAS on diabetes-related outcomes to prioritize causal
associations for intervention trials. Clustering of genetic loci by phenotypic correlations and causal effects will
pinpoint genetic mechanisms of diet that causally influence metabolic disease. We will extend MR to all UK
Biobank outcomes to map comprehensive causal bidirectional relationships with diet.
Overall we will identify novel and multi-ethnic genetic associations with comparable dietary phenotypes across
diverse studies to elucidate the mechanisms of dietary intake and uncover causal relationships between diet,
diabetes, and overall human health. To achieve my goal of becoming an independent investigator in
nutrigenomic and metabolic disease research, I have designed a detailed K99 plan with didactic coursework and
co-mentoring by Drs. Florez, Hirschhorn, and Willett in metabolism, statistical genetics, and nutritional
epidemiology. During the R00 phase, while conducting independent research and continuing to develop research
skills, I will maintain and cultivate collaborations in nutrition and genetics and grow my research program.
