Optimizing minimum spanning tree using stochastic–Variable neighborhood search for efficient clustering of cancer gene data

• Published in: Concurrency and computation Vol. 35; no. 5
• Main Authors: M, Pavithra, RMS, Parvathi
• Format: Journal Article
• Language: English
• Published: Hoboken, USA John Wiley & Sons, Inc 28-02-2023; Wiley Subscription Services, Inc
• Subjects: Algorithms; Clustering; Combinatorial analysis; DNA chips; DNA microarray; Gene expression; Graph theory; Heuristic methods; minimum spanning tree; Optimization; Representations; Search algorithms; stochastic diffusion search; variable neighborhood search algorithm
• ISSN: 1532-0626; 1532-0634
• DOI: 10.1002/cpe.7573

Summary
With the rapid advancement of DNA microarray technologies, researchers in biology can now monitor several gene expression levels in one single experiment that is very useful in detecting or classifying specific types of tumors. The microarray data and its high‐dimensional nature, along with a small set of samples, have presented a new but well‐known phenomenon called the curse of dimensionality. The process of organizing all objects into various groups based on their traits that are either similar or dissimilar is known as clustering. The framework has the foundation of a minimum spanning tree (MST), which is the representation of one set of gene expression data that is multidimensional. The main property of such representations is that every cluster of this expression data will correspond to a subtree in the MST that can convert the problem of multidimensional Clustering to the problem of tree partitioning. A stochastic diffusion search (SDS) has been deployed for clustering. The algorithm was used to be a multi‐agent technique of search and optimization. A variable neighborhood search (VNS) algorithm has been proposed to be a metaheuristic that solves problems of global and combinatorial optimization. The goal of this method was proceeded to get a systematic change to the neighborhood in a local search algorithm. The algorithm will continue to remain the same as an optimal solution that explores the far neighborhoods until better solutions are identified.