Interactive visualization and analysis for gene expression data

Currently, the cDNA and genomic sequence projects are processing at such a rapid rate that more and more gene data become available. New methods are needed to efficiently and effectively analyze and visualize this data. We present a visualization method which maps the samples´ n-dimensional gene vectors into 2-dimensional points. This mapping is effective in keeping correlation coefficient similarity which is the most suitable similarity measure for analyzing the gene expression data. Our analysis method first removes noise genes from the gene expression matrix, then adjusts the weight for each remaining gene. We have integrated our gene analysis algorithm into a visualization tool based on this mapping method. We can use this tool to monitor the analysis procedure, to adjust parameters dynamically, and to evaluate the result of each step. The experiments based on two groups of multiple sclerosis (MS) and treatment data demonstrate the effectiveness of this approach.