Background There are no current disease-modifying treatments for Huntington’s disease (HD). Further analysis of previously curated HD-associated datasets using updated annotations and integration of multiple datasets could highlight therapeutic targets: however, with the increase in size and complexity of both dataset and annotation sources, obtaining biologically useful information is a difficult task.
Aims We investigated methods to characterise modules of co-expressed genes derived using weighted gene co-expression network analysis (WGCNA) in human HD brain. We then applied these modules to uncover parallels between human disease and mouse model phenotypes in HD. We further tested for enrichment of HD genetic modifiers in these modules.
Methods A customised g:Profiler service re-annotated WGCNA modules using human HD, neuropathologically normal, mouse HD models and Alzheimer’s disease datasets. P-values were calculated using cumulative hypergeometric mid-P tests and adjusted using the Benjamini-Hochberg false discovery rate. Jaccard and overlap coefficient algorithms filtered to highlight significant terms and remove redundant ones. Consensus WGCNA networks were built between datasets containing human and mouse samples and datasets associated with different neurodegenerative conditions. Modules were tested for enrichment of the Genetic Modifiers of HD genome wide association study.
Results HD-associated modules from WGCNA performed in human brain expression data were found to be significantly enriched in pathways including regulation of isotype switching which involves recombination and thus DNA cleavage and repair, and mRNA 3’-end processing (p ≤ 0.05).
Conclusions Integration of multiple data sources sheds light on HD biology. Our modifications to g:Profiler increase the statistical validity of the annotations while aiding biological clarity.
Support Cardiff University, School of Biosciences, MRC Centre for Neuropsychiatric Genetics and Genomics
- Network analysis
- Gene annotation