Biclustering of gene expression data using Greedy Randomized Adaptive Search Procedure

Biclustering algorithms belong to a distinct class of clustering algorithms that perform simultaneous clustering of both rows and columns of the gene expression matrix and can be a very useful analysis tool when some genes have multiple functions and experimental conditions are diverse. The biclustering approach thus focuses on finding a subset of the genes and a subset of the experimental conditions that together exhibit coherent behavior. In [1], Cheng and Church have introduced an initial measure called mean squared residue score to evaluate the quality of a bicluster and has become one of the most popular measures to search for biclusters. In this paper, we propose a new method to detect significant biclusters from large microarray datasets. Our method has two major steps. First, high quality bicluster seeds are generated by means of k-means clustering. In the second step, these seeds are grown using a multi-start metaheuristics - Greedy Randomized Adaptive Search Procedure (GRASP). It is an iterative search procedure where each iteration consists of a construction phase followed by a local search procedure. The construction phase of GRASP is essentially a randomized greedy algorithm. Repeated application of the construction procedure yields diverse starting solutions for the local search. Our experiment shows that the GRASP algorithm is efficient and is able to discover coherent biclusters.