A Binary Particle Swarm Optimization Approach for Gene Expression Biclustering Problem

TL;DR

This paper introduces a new binary particle swarm optimization method for gene expression biclustering, using a novel cost function that considers biological relevance, MSR, and bicluster size, showing improved results over existing algorithms.

Contribution

A novel binary particle swarm optimization model with a new cost function for gene expression biclustering, incorporating biological relevance and size considerations.

Findings

Proposed approach outperforms existing algorithms

Effective in identifying biologically relevant biclusters

Validates the use of DCM as a cost function

Abstract

Microarray techniques are widely used in Gene expression analysis. These techniques are based on discovering submatrices of genes that share similar expression patterns across a set of experimental conditions with coherence constraint. Actually, these submatrices are called biclusters and the extraction process is called biclustering. In this paper we present a novel binary particle swarm optimization model for the gene expression biclustering problem. Hence, we apply the binary particle swarm optimization algorithm with a proposed measure, called Discretized Column-based Measure (DCM) as a novel cost function for evaluating biclusters where biological relevance, MSR and the size of the bicluster are considered as evaluation metrics for our results. Results are compared to the existing algorithms and they show the validity of our proposed approach.