A Principal Component Analysis Based Microarray Data Bi-Clustering Method

Microarray data bi-clustering is very helpful for the research on gene regulatory mechanisms analysis. Genes exhibiting similar expression patterns provide useful clues for studying their possible functions. In this paper a novel bicluster detection method is proposed. Compared with the other approaches, biclusters are not detected directly with the whole given experiment data matrix, but are verified with the concatenation of small biclusters which are firstly detected using a conventional clustering method such as K-means and so on so as to making fully use of the rich and powerful existing data clustering methods. By this way, the affect of the high dimensionality of the data is greatly reduced. Since the data within a bicluster is highly correlated with each other, a principal component analysis based efficient verification method is applied to concatenate small biclusers into a larger one. Some experiment results on the simulated data are presented.