A New Two-steps Gene Expression Data Clustering Method

Gene expression data analysis is very important for the research on gene regulatory mechanisms. Genes which exhibit similar patterns are often functionally related. In this paper a novel bicluster detection method is proposed. Its advantage lies in it can not only make use of the traditional data clustering methods, but also form a systemic architecture. The whole processing procedure can be divided into two parts. The first is using one existing clustering method to cluster all the 2-combinations of the data matrix in the direction where the dimensionality is smaller, Then based on the clustering results some binary tables are created. The second part is to verify the concatenated quasi-bicluster. Since the data in the same bicluster is highly correlated with each other, a principal component analysis (PCA) based efficient verification method is applied, which can also work in noisy environment. The whole processing aims at finding all of the possible biclusters from large to small.