Statistical modelling strategies in molecular epidemiology, with an application to attention-deficit hyperactivity disorder
Since 1990s, the technological developments in measuring molecular data have been instrumental in advancing molecular epidemiology. A consequential challenge is to integrate the evidence from multiple molecular datasets for a better understanding of the biological mechanisms underlying complex trait...
| Main Author: | |
|---|---|
| Other Authors: | , , , |
| Format: | Doctoral or Postdoctoral Thesis |
| Language: | unknown |
| Published: |
School of Public Health, Imperial College London
2021
|
| Subjects: | |
| Online Access: | http://hdl.handle.net/10044/1/96270 https://doi.org/10.25560/96270 |
| Summary: | Since 1990s, the technological developments in measuring molecular data have been instrumental in advancing molecular epidemiology. A consequential challenge is to integrate the evidence from multiple molecular datasets for a better understanding of the biological mechanisms underlying complex traits. In this thesis, I applied different statistical modelling approaches to investigate the biological background of attention-deficit/hyperactivity disorder (ADHD), a common neurodevelopmental disorder with an early onset, high persistence and a notable impact on the global burden of disease. I examined the putative impact of exposure to maternal smoking during pregnancy on the risk of ADHD and other adverse outcomes in the offspring related to epigenetic modifications and other molecular changes. The potential causality between ADHD and obesity was analysed using genetically informative methods. The link between systemic chronic inflammation, which can be triggered by smoking or obesity, and common psychiatric outcomes was also investigated. The main dataset used was the Northern Finland Birth Cohort 1986 (N = 6,728 for ADHD symptoms, N = 432 for phenotype and full omics data available), with complementary data from other European cohorts and publicly available summary statistics. I used integrative statistical approaches that leverage evidence from different omics datasets. Regression modelling and Mendelian Randomisation techniques were applied throughout, and a recently published network method based on sparse canonical correlation analysis was also used. The results showed evidence for a long-term impact of intrauterine smoke exposure on offspring DNA methylation, and some indication that DNA methylation mediates the effect of the smoke exposure on offspring later life health outcomes. There was also suggestive bidirectional causality between ADHD and obesity, and evidence for an inflammatory component in the aetiology of psychiatric outcomes. This thesis adds to the literature by a thorough investigation of different omics datasets and integrative statistical approaches applied to ADHD and other psychiatric outcomes. Open Access |
|---|