Comparative Study for Bi-Clustering Algorithms: Historical and Methodological Notes

Abstract

Through subsets of experimental conditions, several current bi-clustering algorithms have been used to evaluate co-expressed genes in data on gene expression. These techniques of data mining are considered essential in the analysis of the microarray data. Such kinds of data analysis often concentrate on the ideal functions of the target relationships between genes and the transcription factor. Bi-clustering approaches look for sample-variable correlations in the form of distinguished data matrix sub matrices. In this paper, we present comparative study for bi-clustering algorithms, which used for numerical experiments are implementing to evaluate large (score and average) of sub matrices (bi-clusters) that found in the original matrix. Two algorithms which has gained importance in recent years in the field of Bi-clustering were discussed and tested within a given two data sets (yeast data set and breast cancer data set). The results suggest that two algorithms (LAS and single bi-cluster) work randomly initialized iterative search procedure that used for the sub matrix and prove that (single bi-cluster) is an effective technique to find maximum score that extracted from ( LAS) algorithm and evaluate large average of original data matrix.