Application of bioinformatics tools for better interpretation of psychiatric GWAS through integration of biological information

Abstract

Psychiatric disorders are common, heritable, often chronic and devastating illnesses who heavily decline quality of life of the patients and their environment. The high genetic correlations across the disorders and their diagnostic criteria reinforce the growing discomfort with the current classification and boost the search for more refined measurements. Genome-wide association studies (GWAS) are a highly successful method for identifying common genetic risk variants underlying common disorders. In psychiatric disorders, the emerging picture suggests contribution from a large number of single-nucleotide polymorphisms (SNPs) of individually small effect sizes as well as rare copy number variants (CNVs) and rare variants discovered by next-generation sequencing. Most of these findings have emerged during the last years through large collaborative efforts which enabled powerful meta-analyses. Nevertheless, individual SNPs and CNVs seem to explain only a minor fraction of the heritable variance for psychiatric disorders. Therefore, the development and correct application of novel bioinformatics methods is necessary to cope with the limitations inherent to GWAS. Biology-informed methods already led to important advances with many discoveries of common, rare and de novo variants that are converging on specific pathways and biological mechanisms. The studies described in this thesis aim to deepen our understanding of psychiatric disorders through the application of novel bioinformatics tools to existing GWAS data sets. We found evidence that schizophrenia-associated loci contribute to the development of bipolar disorder and that the overlapping SNPs converge in pathways previously reported in other psychiatric disorders. We revealed two genes and a pathway significantly associated with borderline personality disorder previously implicated in mental disorders and demonstrated the statistically significant genetic overlap with other psychiatric disorders. We identified two pathways suggesting an involvement of neurodevelopmental processes in the etiology of bipolar disorder. We found that common variants at nine previously reported BD-associated miRNAs do not strongly contribute to the differential responses to lithium treatment in BD. Taken together, these studies show that the application of biology-informed bioinformatic methods enhance the insights gained from GWAS and demonstrate the plethora of methods available nowadays. It is the hope that the progress in understanding the genetic architecture of psychiatric disorders will also help to improve the clinical classification and ultimately yield in better treatment options.