Publikacje

Abstract:

The main objective and contribution of the paper is the application of our original data clustering technique based on generalized self-organizing maps (SOMs) with evolving splitting-merging tree-like structures to a multi-step clustering of colon cancer gene expression data coming from microarray experiments. Such a clustering aims at uncovering a small set of informative genes (i.e., the disease relevant genes) from their overall huge number. Our approach corresponds to the subspace clustering which is highly desirable in microarray data analysis. It is worth emphasizing that our approach works in a fully unsupervised way, i.e., using unlabelled data and without a predefined number of clusters. It is particularly important in the gene clustering of microarray data where the number of gene clusters is unknown in advance. As far as the sample clustering of the colon cancer gene data is concerned, our approach gives better results than those reported in the literature (the alternative methods require, additionally, the cluster number to be defined in advance). The aforementioned experiments are preceded by an outline of our clustering technique and an illustration of its operation with the use of a benchmark data set.

POWRÓT

Strona główna > Publikacje

Uncovering informative genes from colon cancer gene expression data via multi-step clustering based on generalized SOMs with splitting-merging structures