Gene expression data clustering using tree-like SOMs with evolving splitting-merging structures

Abstract

The paper presents an application of our clustering technique using generalized tree-like SOMs with evolving splitting-merging structures to complex clustering tasks, including, in particular, the sample-based and gene-based clustering of the Lymphoma human cancer microarray data set. It is worth emphasizing that our approach works in a fully unsupervised way, i.e., using unlabelled data and without the necessity to predefine the number of clusters. It is particularly important in the gene-based clustering of microarray data for which the number of gene clusters is unknown in advance. In the sample-based clustering of the Lymphoma data set, our approach gives better results than those reported in the literature (some of alternative methods require, additionally, the cluster number to be defined in advance). In the gene-based clustering of the considered microarray data, out approach generates clusters that are easily divisible into subclusters related to particular sample classes. In some way, it corresponds to subspace clustering that is highly desirable in microarray data analysis.