Identification of coexpressed gene modules across multiple brain diseases by a biclustering analysis on integrated gene expression data

Abstract

It has been reported that several brain diseases could share symptoms at clinical level, suggesting the necessity and possibility to develop therapeutics. In this paper, we carried out an integrated gene expression analysis on several microarray datasets of neurodegenerative diseases and psychiatric disorders to discover the uniqueness and commonness in their molecular basis. First, we selected and combined three sets of microarray data including eight brain diseases. Second, we applied a correlation-based biclustering approach, BICLIC [1], to efficiently identify coexpressed gene modules that are correlated in individual or multiple combinations of brain diseases. Third, Gene ontology-based functional enrichment analysis is performed to analyze functional characteristics of the identified cross-disease or and disease-specific modules. In this approach, we could examine various sets of correlated genes significantly in both single and multiple diseases. As a result, in total, 4,307 coexpressed gene modules were turned out to be common to two or more of brain diseases. Among them, eight modules having different combinations of total 16 genes were involved correlatively in more than seven brain diseases. The functional analysis showed that the multi-disease specific modules were more associated to higher brain functions like cognitive functions than single disease specific modules. The results in this study provide valuable resources to further investigate the key molecular players affecting on brain diseases in both transnosological or disease specific manner.