A Projection and Search Algorithm for the Bicluster Problem

The advent of microarray technologies has made the experimental study of gene expression faster and more efficient. It is an important task to analyze the gene expression data acquired from the microarray experiments. Biclustering is a popular method to search a subset of genes which exhibit similar expression patterns along a subset of conditions in the gene expression matrix. In this paper, we propose a projection and search algorithm (PSA) for finding biclusters in the gene expression matrix. The algorithm adopts projection method to find a raw submatrix, then refines and extends the raw submatrix into a bicluster using greedy algorithms. We test the algorithm on simulation data and real data. The test results show that the algorithm has good performance on both simulation data and real data. Additionally, compared with other methods, PSA isnpsilat sensitive to noise level.