Data Clustering Using Particle Swarm Optimization for Pairwise Microarray Bioinformatics Data

This paper proposed an approach to using particle swarm optimization to cluster the pairwise of similarity data from pathway enrichment analysis of microarray experiment. The data from this experiment are illustrated in one-mode NxN similarity matrix. Such data will describe the pairwise relations between objects which are pathway enriched by significant information. The PSO algorithm tends to find the best of grouping that maximized the criterion function I
2
which is the square-root of the dot product of the pairwise similarity between all the document in particular cluster. Comparison the cluster analysis with hierarchical clustering, default method in pathfindR package with k suggestion and traditional K-means clustering. Our approach can rearrange the number of k to reach the best grouping that maximized criterion function. Likewise, our algorithm results come to the same conclusion with K-means clustering and hierarchical clustering with Cohen’s kappa coefficient is 0.6627 and 0.7956, respectively. Our method allows us to see more varieties of the clustering results. This will be beneficial for future researchers’ use thanks to its better exploration of results.