Mapping binary (On/Off) gene expression for pathway & tissue analysis

Abstract

Microarray data contain information about the level of expression of genes that can be informative of changes taking place in cells. This information has been widely used to study the changes in gene expression between normal and cancer cells. Gene expression has been used as a biomarker predictive of the progression of a disease and to identify drug targets specifically expressed in cancer tissue. Although the level of expression of a gene can change by multiple folds across tissues, the simplest information (on/off status) about a gene is whether it is or is not expressed in a given tissue. In this study we propose to take advantage of a large set of tissue specific gene expression data to study the profile of gene expression in pathways and tissues. To perform this analysis, we leverage and improve a method that computes the on/off status of a gene from their level of expression. The percent of genes with on status in a given tissue was selected to summarize across bio specimen. The gene state method was applied to sets of tissue specific expression microarray extracted from the GEO database. We then studied the profile of on/off state of genes in KEGG pathways across several tissues. The data were then used to calculate a distance between gene sets. Using all genes, a distance could be calculated between normal and cancer tissue, as well as pairwise comparisons between each tissue type. The gene sets were then narrowed and selected based on pathway annotation from KEGG. This demonstrated an ability to identify known cancer pathways based on their gene signature distances. The results affirm known cancer pathways by calculating relative distances.