Date of Award

Fall 9-5-2025

Degree Type

Thesis

Degree Name

Master of Science (MS)

Department

Bioinformatics & Computational Biology

First Advisor

Heather Wheeler

Abstract

Given genetic and environmental diversity, genetic prediction based on African- ancestries individuals in the US or UK may not transfer well to other African diaspora regions. The Modeling the Epidemiological Transition Study (METS) is a longitudinal study of cardiometabolic health in five African-ancestry populations. We performed genome-wide genotyping in METS participants from the US, Jamaica, South Africa, and Ghana. We aimed to explore how genetic models for the transcriptome and body mass index (BMI) built using US- based African-ancestries populations perform in these four diverse populations. A subset of METS participants also had whole blood RNA-Seq data. For these, we predicted gene expression from genotypes using published African-ancestries transcriptome models, which were built in the US-based TOPMed MESA study. Next, we used the three largest populations in All of Us to build cross-validated polygenic risk score (PRS) models for BMI with PRS-CSx, a Bayesian method that combines genetic effects across populations via a continuous shrinkage prior. We then tested our PRS models in METS participants. PRS performance varies between these four groups, roughly following the trend in their countries’ human development index levels, suggesting the influence of environmental differences in genetic prediction of BMI. Transcriptome prediction performance also varies significantly between the US and Ghana. Overall this work supports the importance of global representation in genetics research for more accurate healthcare.

Download

Share

COinS