Conclusion
That the interactions between the trifecta of human genetics, brain functioning and mental health are complex is not a statement that needed reiterating. Instead, the work discussed here showed new avenues by which these interactions can be understood. Among these avenues are the first application of multivariate GWAS implemented through MOSTest on fMRI data, using ICA combined with GWAS to show that even though phenotypes relevant to mental health might be statistically independent they nonetheless share genetic architecture, and using a combination of genetic mapping tools to show that the associations between the independent phenotypes, psychiatric disorders, and the functional connectome are linked to biological processes previously associated with the brain. These factors combined advance our understanding on how mental health, genetics, and brain functioning are linked.
The field of psychiatric genetics suffers from some inherent issues that need to be addressed. That being said, the developments in novel methods such as multivariate GWAS implemented through MOSTest1, analysis on the shared genetic architecture implemented through pleioFDR2 and MiXeR3, mapping of genetic loci implemented through FUMA4, and a host of other tools providing that allow researchers to map the genetic findings to biological processes. Through a combination of further development and aligning of these tools, combined with improvements in available data, the wholistic approach of analysing the brain in concert with genetics and mental health will provide key insights into the brain’s functioning and dysfunction in psychiatry.
These conclusions are somewhat confounded by the statistical limitations of fMRI in this context. Although MOSTest provides the necessary boost in effective sample size to make the analyses discussed here possible, the oblique nature of fMRI data is still the bottleneck for these types of analyses despite the use of large biobanks such as the UKB.
In the work under evaluation in this thesis made extensive use of abstractions of concepts when it came to mental health (i.e. through the ICA approach of measuring mental health phenotypes), genetics (through MOSTest), and the brain. Useful insights beyond fundamental research are difficult to derive from these abstractions. In order to further elucidate shared etiology of poor mental health outcomes and psychiatry, research should inevitably return to specificity when the tools and datasets allow for it. Currently, to reliably pick up signals as faint as genetic associations with fMRI various levels of abstractions are necessary to be able to draw reliable conslusions on the nature of the subject under study.
More importantly, in order to make this research relevant beyond the minor advances in biological knowledge and methodological insights, improvements in the data used for studies in the psychiatric genetics field are required. More emphasis is required on collecting data from marginalized groups and the Global South. Only when proper diversity in the data available is ensured can this field start to build tools that may one day be both clinically relevant and generalizable. The path ahead for psychiatric genetics is well-lit by the work that came before, but that also means the obstacles in the way are clearly visible.