There is a wealth of information for a multitude of human diseases in public repositories of gene expression microarray datasets. These data can be utilized to generate new hypotheses that will allow better diagnosis, categorization and treatment of human diseases.
We provide a framework to utilize a hitherto under utilized higher order complexity in gene expression data that elucidates the regulatory changes characteristic of human disease. We have applied this framework to the deadly brain cancer glioblastoma multiforme and demonstrate that this approach is capable of identifying clusters of functionally relevant co-regulated genes.
The benefit of our approach is that through integration of de novo motif detection we are able to identify cis-regulatory factors driving disease related processes. By integrating clinical traits with this new found information we will be able to provide new insights into human disease, which can then be tested and used to improve prevention, diagnosis, prognosis, sub-classification and treatment of human disease.
|Global gene expression profiles are converted to sets of co-regulated genes that can provide novel insights into human disease. Sets of co-regulated genes are regulated by three different types of factors: environmental factors, cis-regulatory factors, and clinical traits. Using correlation we can relate a set of co-regulated genes to a clinical trait. Through network inference, ontological overlap analyses, and motif comparisons this approach can provide novel insights into human disease.|