Parallel Coordinate Plots of Gene Expression Data Differentiated by Phenotype

• Activity Number: 111
• Type: Contributed
• Date/Time: Monday, August 9, 2004 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistical Graphics
• Author(s): Kellie J. Archer*+ and Viswanathan Ramakrishnan
• Companies: Virginia Commonwealth University and Virginia Commonwealth University
• Keywords: microarray ; gene expression ; parallel coordinates

Abstract

While methods for visualizing microarray data associated with an analytical results exist, such as displaying a heatmap with associated dendrograms from hierarchical cluster analysis, in general, visualization methods for raw microarray data are not readily available due to the inherent multidimensionality. Recently, plots of gene expression data that incorporate biological information have become available. One example is the cumulative gene expression plot for a selected chromosome suggested by Gentleman (2003). What is ultimately of interest in the clinical use of microarrays is the measurement of differences in signal intensities between different phenotypes. A modification of Gentlemans cumulative expression plot by chromosome, where absolute expression for each gene is plotted via parallel coordinates, is presented. In addition, color is added to permit visualization of difference due to phenotype. These plots may aid in determining regions of deleted chromatoid material characteristic of a phenotype and may be more sensitive than traditional cytogenetic analyses. The parallel coordinate plot will be demonstrated using 31 breast cancer samples hybridized to HG-133A chips.