Applications of Network Clustering in Natural Product Research

Abstract

Clustering is an important issue in data analysis in diverse fields of study. There is no universal definition of a cluster. Consequently, there are many different definitions and various methods. The major purpose of clustering is finding cohesive groups. Clustering can be applied to multivariate data or also to networks/graphs. In this article, we are going to discuss some applications of graph clustering in natural product research. For global-level understanding, initially the elements of a system are connected based on their mutual relations and a network is formed on this basis. For example, networks at the molecular level are constructed to understand and explain the cell as a system. Numerous studies constructed suitable networks for understanding systems or subsystems within a species or across a number of species. This world is cohabitated by humans and many other species and the understanding of the interactions at the molecular level among all the species is important for healthy and sustainable living for humans and other organisms. Natural products produced by plants and other species play important roles in human healthcare and living. Therefore, advancing natural product research is useful in many sense for guiding human civilization to prosperity. There can be many applications of network algorithms particularly clustering algorithms in natural product research. It is expected that structurally similar molecules produce similar MS/MS spectra. Therefore, molecular networks are used to organize MS/MS spectra into groups based on similarities in their fragmentation patterns. Molecular networking offers a fast and effective approach to compare metabolic profiles among strains without employing complicated data mining. Molecular Networking based on MS/MS data has been applied for the discovery of biosynthetic gene clusters and their products from Salinispora species. It has been used for the global visualization of the molecules produced by single or multiple organisms, to discover new natural products from Streptomyces coelicolor. OrthoMCL is a method for constructing orthologous groups within single or across a number of eukaryotic species, using a Markov Cluster algorithm, which is widely known as MCL algorithm. Doroghazi et al. presented a roadmap for natural product discovery based on large-scale genomics and metabolomics involving OrthoMCL. In this book chapter the MCL algorithm will be discussed somewhat in detail because it has been utilized not only in natural product research but also in many other omics studies. Altaf-Ul-Amin et al developed graph clustering algorithms DPClus and DPClusO which have been utilized in various research such as protein–protein intercation, metabolomic correlation-network, transcriptome analysis of Oryza Sativa, molecular marker and disease pathways, classification of ions into metabolite-derivative groups etc. Recently a biclustering algorithm called BiClusO based on the DPClusO algorithm has been developed. In this article, the DPClusO and BiClusO algorithms will be presented somewhat in detail. Also, some of the applications of DPClus, DPClusO and BiClusO in natural product related research will be explained elaborately.