A parallel Clustering algorithm based on minimum spanning tree for microarrays data analysis

Abstract

Clustering is partitioning a set of observation into groups called clusters, where the observation in the same group has a common characteristic. One of the best known algorithms for solving the microarrays data clustering problem using minimum spanning tree (MST) is CLUMP algorithm (Clustering algorithm through MST in Parallel) which identifies a dense clusters in a noisy background. The MST construction phase of the CLUMP is the time consuming phase. This paper presents an improved version of CLUMP algorithm called iCLUMP (improved Clustering algorithm through MST in Parallel). iCLUMP enhances the speedup of MST construction using the cover tree data structure. The implementation shows that iCLUMP is efficient than CLUMP in terms of complexity and runtime.