Skip to Main Content
Using the DNA microarray technology, biologists have thousands of array data available. Discovering the function relations between genes and their involvements in biological processes depends on the ability to efficiently process and quantitatively analyze large amounts of array data. Clustering algorithms are among the popular tools that can be used to help biologists achieve their goals. Although some existing research projects employed clustering algorithms on biological data, none of them has examined the Escherichia coli (E. coli) gene expression data. This paper proposes a clustering algorithm called Multilayer Adjusted Tree Organizing Map (MA TOM) to analyze the E. coli gene expression data. In a semi-supervised manner, MATOM constructs a multilayer map, and at the same time, removes noise data in the previously trained maps in order to improve the training process. This paper then presents the clustering results produced by MATOM and other existing clustering algorithms using the E. coli gene expression data, and a new evaluation method to assess them. The results show that MATOM performs the best in terms of percentage of genes that are clustered correctly.