Main findings

All original analyses contained in this thesis used data from the UKB. The UKB is a resource that aims to sample variables related to numerous aspects of health in a general population. This has the distinct advantage that participants are not recruited with a particular focus on diagnosis or condition, making the dataset usable for a wide range of topics. However, the UKB is not immune to selection bias and participants showed on average lower obesity rates, alcohol intake, and were generally healthier than peers in the general population of the same age¹. Nonetheless there was sufficient variance in the self-report mental health questionnaire for the ICA to construct 13 meaningful components. However, the selection bias discussed above also affects the genetic discovery in this sample². This unfortunately means that genetic findings from studies using this databank do not necessarily translate fully to the populations outside of the populations sampled in the UKB.

A GWAS on risk factors such as sexual abuse, traumatic experiences, and emotional abuse likely reflects shared effects underlying these factors rather than the risk factors itself. For example, from a data collection perspective the variance driving the associations may instead reflect interpersonal variance in answering the self-report questionnaire. Some individuals are perhaps more likely to honestly report being subject to sexual or emotional abuse than others^3–5. Interpersonal variation in answering the self-report questionnaire may be a latent factor influencing the genetic profiles of some of these components. Additionally, something that was not part of our analyses, there are complex interactions between SES, race, genetics, and exposure to life stressors^6–11. These factors can all influence someone’s risk of exposure to traumatic life events.

Through conjFDR we identified a number of shared loci between the brain functional connectome and a number of psychiatric disorders and profiles reflecting various aspects of mental health generally. These associations between the brain functional connectome and psychiatric disorders further highlight the complex interplay between the brain, genetics, and mental health. Though the connectome has been implicated in psychiatric disorders before¹², the novelty here lies in the number of novel overlapping loci with the various psychiatric disorders, and our ability to link the associated genes to relevant biological processes. We achieved this by deploying a multivariate analysis which boosted the effective sample size of the GWAS and increased the discovery power of the included fMRI phenotypes. These phenotypes often suffer from lack of power due to retained noise, lack of granularity, insufficient sample size, class imbalance, and a host of other issues. While a number of the lead SNPs discovered were also present in some univariate measures, the aggregated phenotype in MOSTest provides more robust metrics that more accurately account for the multiple testing problem while providing a more global measure of functional connectivity than the collection of granular univariate measures would.

Among these identified biological processes were for example synaptogenesis, which provides further support for the common hypothesis that psychiatric disorders, and in particular SCZ, are hallmarked by brain dysconnectivity¹⁴. While the dysconnectivity hypothesis comes with certain limitations that also apply in the context of the current research, our results do support its main assertion from a genetics perspective.