Dissecting causal relationships and molecular mechanisms in disease using genetic risk profiles

Abstract

Polygenic scores (PGS) have shown great potential for predicting disease risk and stratifying individuals with elevated genetic risks for early screening and intervention in the general population. PGS can also be integrated with epidemiological and statistical genetics methods to help elucidate the causality of disease-associated molecular features and lifestyle risk factors.
In this study, I have examined the use of PGS in three epidemiological scenarios, using molecular traits, anthropometric measures, and cardiometabolic outcomes as examples:
I. In Chapter 3, I evaluated a method called block jackknife resampled Mendelian randomization in simulated and applied settings. This technique uses jackknife-resampled PGS as the genetic instrument in individual-level Mendelian randomization. I found that this method generates accurate effect estimates whilst alleviating the biases caused by participant overlap when individual-level data is from a single source.
II. In Chapter 4, I generated a comprehensive atlas of the correlations between 249 metabolic traits and PGSs of 125 complex traits or diseases in the UK Biobank. I used two examples from each side of the association analyses, namely glycoprotein acetyls and coronary heart disease (CHD), to illustrate the utility of this valuable resource.
III. In Chapter 5, I used the PGS for height to explore the associations between early-life exposures and adult CHD. Standing height is largely influenced by infanthood and childhood exposures. PGS-predicted height and PGS-adjusted height residuals are enriched for the genetic and environmental contributors to adult height. Comparisons between their associations and CHD risks highlight the significant contribution of early-life exposure to adult-onset disease.
This thesis explores the versatile applications of PGS to understanding potentially causal relationships in epidemiology. Findings highlight the potential of PGS to enhance method development, unravel molecular causality and uncover the impact of early-life factors on adult health outcomes, providing valuable insights for translational research and public health interventions.

Date of Award	12 Jun 2024
Original language	English
Awarding Institution	University of Bristol
Supervisor	Tom G Richardson (Supervisor), Tom R Gaunt (Supervisor) & George Davey Smith (Supervisor)