Selection of Pathway Markers for Cancer Using Collaborative Binary Multi-Swarm Optimization

Selection of Pathway Markers for Cancer Using Collaborative Binary Multi-Swarm Optimization

Prativa Agarwalla (Heritage Institute of Technology, India) and Sumitra Mukhopadhyay (Institute of Radiophysics and Electronics, India)
Copyright: © 2018 |Pages: 27
DOI: 10.4018/978-1-5225-2607-0.ch014


Pathway information for cancer detection helps to find co-regulated gene groups whose collective expression is strongly associated with cancer development. In this paper, a collaborative multi-swarm binary particle swarm optimization (MS-BPSO) based gene selection technique is proposed that outperforms to identify the pathway marker genes. We have compared our proposed method with various statistical and pathway based gene selection techniques for different popular cancer datasets as well as a detailed comparative study is illustrated using different meta-heuristic algorithms like binary coded particle swarm optimization (BPSO), binary coded differential evolution (BDE), binary coded artificial bee colony (BABC) and genetic algorithm (GA). Experimental results show that the proposed MS-BPSO based method performs significantly better and the improved multi swarm concept generates a good subset of pathway markers which provides more effective insight to the gene-disease association with high accuracy and reliability.
Chapter Preview


Genes control the different functioning of a cell, like growth, division, death etc. When the normal profile of a gene is changed or damaged, it causes the abnormal behavior of the cell and we, in generic sense, call it as cancer. Cancer is nothing but out-of-control cell growth due to change in the expression profile of genes. Advancement of microarray technology has made the genomic study more fast and efficient by analysing the expression of thousands of genes in a single chip (Zhang, 2008). But, the huge dimension of the gene expression data leads to statistical and analytical challenges to identify differentially expressed genes in different classes for the study of their effect on diseases. So the selection of the most relevant genes is very essential for the proper medical diagnosis as well as for drug target prediction and in this context, different aspects of big data processing and analysis come into play. Big data analysis is one of the very popular and recent day technologies, used for examining large dataset. It helps to reveal hidden pattern and correlations information of the data which can be used in the field of computational biology to analyze huge biological data for extracting the relevant information and to enrich the knowledge related to the biological system. While exploring small number of significant genes participating in a tumour progression, it has been observed that those genes are functioning similar and work as a group to form a certain cancer. The set of genes having identical biological functioning is known as a pathway. To understand the biological functioning of those groups of genes, involved in tumour progression and the phonotypical changes at the pathway level is a very interesting research topic now a day. But, the dataset contains high amount of noise and overlapping samples which decreases classification accuracy and the identification of dominant genes related to the disease suffers. Proper and efficient identification of those differentially expressed genes at the pathway level is very crucial for the treatment of the prior disease related to the pathway.

Complete Chapter List

Search this Book: