Weighted gene co-expression network analysis applied to head and neck squamous cell carcinoma data

Abstract

Microarray technology has made possible the simultaneous monitoring of the expression levels of thousands of genes under multiple disease states. Due to the high complexity of the obtained data the use of computational methods for extracting biological evidences is still a major issue. In this work we address this problem by adjusting the use of gene coexpression networks to analyze a Head and Neck Squamous Cell Carcinoma (HNSCC) dataset. The proposed method applies hierarchic clustering to identify gene modules using the topological overlap dissimilarity measure after, defining a gene co-expression similarity, defining a family of adjacency functions and calculating their parameters. This method calculates the eigengenes of each module to define a network of modules and the correlation between the eigengenes and the risk factors, identifying modules of genes where those are more expressed and associating these concepts to gene ontology functional terms. The preliminary results described in this paper contribute to reveal the molecular mechanisms associated with HNSCC and the contribution of experimental factors types like differentiation, alcohol use, sex, age, tumor site, smoking pack years and race.