Imaging genetics

Genetic analysis of the brain at a direct, molecular level is complicated due to the protected location of the brain in the skull and its fragile state. Surgical interventions are highly invasive and carry a high degree of risk and are in a vast majority of studies not suitable as a research procedure. Instead, researchers resort to the much less invasive approach of scanning the brain using either MRI, EEG, MEG, X-ray, CT, or PET scans. Each of these methods have been applied in psychiatry with varying levels of success^1,2. Although each of these approaches comes with its own set of advantages and drawbacks in terms of invasiveness, time required, cost, and resolution, for its flexibility and the ability to keep scanning times low MRI has become one of the major tools researchers use to study the brain^2,3. Like genetics, most MRI findings in psychiatry point towards a large number of alterations each with a small effect size⁴.

MRI research has implicated brain changes in both the structural and functional domain in psychiatry^4,5. There is a large body of work showing structural brain changes in for instance SCZ^6–9, BIP^10,11, MDD^12,13, and ASD^14,15. These widespread alterations also extent to fMRI modalities where studies have identified changes in functional connectivity in SCZ^16–18, BIP^19–21, MDD^22,23, ASD^24,25, ADHD^26,27, PTSD^28,29, and ANX^30,31, among others.

As imaging studies have made use of increasingly larger sample sizes (primarily through large biobanks and consortia), this has created an opportunity to consider the genetic underpinnings of the brain imaging phenotypes. The opportunities of study here are limited only by the phenotypes extracted by for instance MRI and apply a GWAS to each feature in the modalities³². Studies running genetic associations on a large number of features from different modalities show widespread heritability across imaging modalities and features³².

While these broad analyses may reveal genetic overlap between these imaging features, this may in part be due to the similarity in the features under investigations. When all features in a modality reflect brain diffusion, one would expect there to be a significant amount of correlation. Therefore, with the rise of these large-scale imaging genetic studies the need arose to adequately correct for these inherent correlatations. The traditional methods of correcting using the Bonferroni- or FDR \(\alpha\) correction can in these cases be unnecessarily conservative. The novel methods implemented in the works included in this thesis address a number of these limitations and allow for improved discovery and introduce novel approaches to further biological insights into genetic features of mental health and the brain.