Journal article Open Access
Background: Despitehighheritability,thegeneticvariantsinfluencingbipolardisorder(BD)susceptibilityremainlargelyunknown.Low statistical power to detect the small effect-size alleles believed to underlie much of the genetic risk and possible heterogeneity between cohorts are an increasing concern. Integrative biology approaches might offer advantages over genetic analysis alone by combining different genomic datasets at the higher level of biological processes rather than the level of specific genetic variants or genes. We employed this strategy to identify biological processes involved in BD etiopathology.
Method: Three genome-wide association studies and a brain gene-expression study were combined with the Human Protein Reference Database protein–protein interaction network data. We used bioinformatic analysis to search for biological networks with evidence of association on the basis of enrichment among both genetic and differential-expression associations with BD.
Results: Weidentifiedassociationwithgenenetworksinvolvedintransmissionofnerveimpulse,Wnt,andNotchsignaling.Threefeatures stand out among these genes: 1) they localized to the human postsynaptic density, which is crucial for neuronal function; 2) their mouse knockouts present altered behavioral phenotypes; and 3) some are known targets of the pharmacological treatments for BD.
Conclusions: Genetic and gene-expression associations of BD cluster in discrete regions of the protein–protein interaction network. We found replicated evidence for association for networks involving several interlinked signaling pathways. These genes are promising candidates to generate animal models and pharmacological interventions. Our results demonstrate the potential advantage of integrative biology analyses of BD datasets.