Consensus Clustering for Microarray Gene Expression Data

Cluster analysis in microarray gene expression studies is used to find groups of correlated and co-regulated genes. Several clustering algorithms are available in the literature. However no single algorithm is optimal for data generated under different technological platforms and experimental conditions. It is possible to combine several clustering methods and solutions using an ensemble approach. The method also known as consensus clustering is used here to examine the robustness of cluster solutions from several different algorithms. The method proposed here also is useful for estimating the number of clusters in a dataset. Here we examine the properties of consensus clustering using real and simulated datasets.