Clustering Genes, Tissues, Cells and Bioactive Chemicals by Sphere SOM

Abstract

The technology of high throughput screening is nowadays widely available in life science especially in the fields of molecular diagnosis and drug discovery. This is due to the establishment of microarray procedure, which deals with huge number of genes at one time. Cluster analysis is usually performed on the results of DNA microarray experiments. However, the routine procedure of data mining dealing with the huge number of signal information obtained from microarray is not fixed yet. We can find various way of clustering in hierarchical and non-hierarchical methods, which are applied for the analyses. The most popular ones appear non-hierarchical clustering such as k-means (MacQueen, 1965), partitioning around medoids (Kaufman and Rousseeuw, 1990) and cluster affinity search technique (Ben-Dor et al., 1999). We have employed spherical self organizing map (sSOM), which is also a non-hierarchical clustering, to cluster genes by gene expression profiles of cells and tissues (Tuoya et al., 2008). Analyzing various types of carcinoma cells and normal tissues, we could find interesting cell surface molecules, which should serve as the molecular markers. This procedure, which we are demonstrating, is rather new to the data analyses of gene clustering from the gene expression profiles obtained from DNA microarray technique. Flexible arrangement of the data obtained allows us to cluster cells and tissues as well as genes to find definitely fantastic direction of further advancement of study. Furthermore, we applied sSOM to classify bioactive chemical compounds by their mechanism of action (MOA), which should enable us virtual screening in silico (Reddy et al., 2007). In recent years, many intriguing methods for virtual screening have been developed in this field (Gasteiger et al., 2003; Melville et al., 2009). Especially, ligand-based method is suitable for selecting drug candidates from enormous compounds library because it simply requires computational resources, which are less expensive. Although SOM has partly been used as a tool of ligand-based methods to classify compounds by their properties in chemoinformatics, spherical SOM has not been used in chemoinformatics to the best of our knowledge (Brustle et al., 2002; Schneider & Nettekoven, 2003; Schneider & Schneider, 2003; Wang et al., 2005; Kaiser et al., 2007; Renner et al., 2007; Li & Gramatica, 2010). We propose here the extended application of sSOM to classify bioactive compounds by their MOA together with their structural information. In the future this procedure should