MSVD-MOEB algorithm applied to cancer gene expression data

Cluster analysis of cancer gene expression data can provide bases for the early diagnosis of cancer and accurate classification of cancer subtypes. Aiming at the characteristics of cancer gene expression data, an algorithm which is called MSVDMOEB (Modular Singular Value Decomposition Multi-Objective Evolutionary Biclustering) is proposed. MSVD-MOEB algorithm applies the singular value matrix to the gene expression matrix after its decomposition and improvement to obtain a meaningful biclustering, then uses multi-objective evolutionary algorithm to perform the global optimization; and finally utilizes the cluster expansion and merging algorithms to find the maximized biclustering. Matlab experiment result shows that MSVD-MOEB algorithm improves the calculating speed of the algorithm and has high accuracy of clustering.