Finding distinct biclusters from background in gene expression matrices



Zhengpeng Wu 1, $, Jiangni Ao 1, $, Xuegong Zhang1, *



1Bioinformatics Division, TNLIST and Department of Automation, Tsinghua University, Beijing 100084, China; $These authors contributed equally to this work


Email; * Corresponding author


Article Type

Prediction Model



received December 15, 2007; accepted December 31, 2007; published online December 30, 2007



Biclustering, or the discovery of subsets of samples and genes that are homogeneous and distinct from the background, has become an important technique in analyzing current microarray datasets. Most existing biclustering methods define a bicluster type as a fixed (predefined) pattern and then trying to get results in some searching process. In this work, we propose a novel method for finding biclusters or 2-dimensional patterns that are significantly distinct from the background without the need for pre-defining a pattern within the bicluster. The method named Distinct 2-Dimensional Pattern Finder (D2D) is composed of an iterative reordering step of the rows and columns in the matrix using a new similarity measure, and a flexible scanning-and-growing step to identify the biclusters. Experiments on a large variety of simulation data show that the method works consistently well under different conditions, whereas the existing methods compared may work well under some certain conditions but fail under some other conditions. The impact of noise levels, overlapping degrees between clusters and different setting of parameters were also investigated, which indicated that the D2D method is robust against these factors. The proposed D2D method can efficiently discover many different types of biclusters given that they have distinctive features from the background. The computer program is available upon request.



gene expression matrices; simulation; biclusters; Distinct 2-Dimensional (D2D); noise


Wu, et al., Bioinformation 2(5): 207-215 (2007)


Edited by

A. M. Khan, T. W. Tan & S. Ranganathan






Biomedical Informatics



This is an Open Access article which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. This is distributed under the terms of the Creative Commons Attribution License.