Semantic similarity based feature extraction from microarray expression data

Previous studies have proven that it is feasible to build sample classifiers using gene expression profiles. To build an effective sample classifier, dimension reduction process is necessary since classic pattern recognition algorithms do not work well in high dimensional space. In this paper, we present a novel feature extraction algorithm by integrating microarray expression data with Gene Ontology (GO). Applying semantic similarity measures, we identify the groups of genes, called virtual genes, which potentially interact with each other for a biological function. The correlation in expressions of virtual genes is used to classify samples. For colon cancer data, this approach significantly improved the classification accuracy by more than 10%.