Discovery of Overlapping Pattern Biclusters from Gene Expression Data using Hash based PSO

Most of the biclustering algorithms believe on grouping the data elements on the basis of distance metric between objects and conditions, where as in real data, strong correlations may exist among a set of data elements and conditions even if they are far apart from the measuring candidate which can be identified in the form of similar patterns such as scaling and shifting patterns. Most of the biclustering methods usually perform their tasks under the assumption that each gene belongs to only one bicluster. But, depending on the experimental conditions being investigated, each gene may have similar expression pattern with different genes in different biclusters and they can, therefore, belong to more than one bicluster. In this paper, an efficient model has been proposed which captures the coherent behaviour among the data elements measuring the co-expression among data elements in the hybridized framework of hashing and Particle Swarm Optimization (PSO) and also discovers overlapping biclusters which can lead to discovery of the great biological complexity.