Microarray technology allows the measurement of expression levels of thousands of genes simultaneously and is expected to contribute significantly to advances in fundamental questions of biology and medicine. While microarrays monitor thousands of genes, there is a lot of evidence that only a few underlying signature components of gene subsets account for nearly all of the outcome variation. Here, methodology for revealing these predictive gene clusters in microarray data is presented. For this task, we focus on supervised algorithms, defined as clustering techniques which utilize external information about the response variables for grouping the explanatory variables (genes). In studies where external response variables are available, our approach is often more effective than unsupervised techniques such as hierarchical clustering.