Network Analysis of Genome-wide Association Studies of Psychiatric Disorders

Abstract

Network analysis is a tool in computational genomics for identifying a common bio- logical pathway linking different genes of significance to a given disease. The technique is applied to genome-wide association studies (GWAS), which associate genetic variants with a disease based on frequency. GWAS have long been used on psychiatric disorders, but have had limited success due to the divergent causes of mental illness and the frequently inadequate size of studies. However, recent GWAS have achieved the sample sizes necessary to identify genes associated with depression as well as other psychiatric disorders. Depression is thought to be a mutationally heterogeneous and polygenic disorder. The large number of mutations that are associated with, but not causally linked to depression suggests that network analysis may be a technique that would help uncover the relationship of different variants influencing depression and other psychiatric diseases. We used the GWAS associations from a myriad of mental disorders including depression, bipolar disorder, and schizophrenia, as well as various other nonmental diseases, in order to examine the relation- ship between genes in the context of protein-protein interaction networks and tissue-specific functional networks. We analyzed different attributes of interest within these networks such as graph density, component size, and cross-disorder gene overlap. Ultimately, we were able to provide an overview of the topology of the disorders we studied, suggest additional at- tributes for consideration in network analysis of psychiatric disorders, and use brain-specific tissue networks to provide clear evidence for the role of genetic alternation along pathways of the brain in psychiatric disorders.