Interpretation and improvement of the current genetic epidemiology methodology for schizophrenia

Abstract

Recent technological advancements have allowed for the development of new methodologies in the investigation of the genetic epidemiology of schizophrenia. Two of the most prominent methods in that field are the Polygenic Risk Scores (PRS) and the Genetic Restricted Maximum Likelihood (GREML) approach. Both methods analyse Single Nucleotide Polymorphism (SNP) data to calculate heritability estimates for polygenic traits. Three separate studies were carried out in the context of this thesis to investigate the means through which these methods operate and devise ways to optimise their function, as well as, improve their interpretability, in the context of schizophrenia research. The first and second study focused on PRS, with the former attempting to use experimentally derived information to improve the interpretability of PRS results, in the context of translating them into meaningful information for the genetic architecture of schizophrenia, while in the latter, three prominent means of applying PRS were compared in extensive simulated scenarios. In the third study, an application of GREML on a population cohort was used as a means to investigate the possibility that this method might be affected by hidden population substructure. Furthermore, a comparison between GREML and GREML-IBD which takes into account rare variants to calculate heritability estimates is made. This thesis highlighted potential methodological limitations of two of the most commonly used approaches in schizophrenia research and through their implementation on both population and clinical-based samples proposed novel means of improving them.