Abstract
The past decade has seen a rapid growth in the application of mathematical and computational tools for extracting insight from biological networks, and of particular interest here, visualising the community structure within such networks. Clustering approaches have proven useful methods to uncover structural and functional sub-groups from within protein interaction networks. However many commonly used clustering methods for identifying functionally relevant substructures within molecular networks do not perform well with increasing network sizes. We tested the performance of algorithms in terms of their ability to identify functionally relevant sub-clusters within networks of varying size as well as computational performance. Our studies suggest many algorithms perform well on smaller networks but fail to scale with network size. A Spectral based Modularity clustering algorithm, with a fine-tuning step, provided both scalability and improved identification of clusters enriched for functional annotation (e.g. disease) in real proteomic interaction datasets.
Highlights
The detection and analysis of community structure in networks has received considerable attention in recent years [1,2,3]
The package clusterCons [20] has been used in conjunction with the suite, to build a consensus matrix from which to test the robustness of discovered communities, and proteins found inside each community
We find our implementations reproduce results found in the MASC network study, whilst our Spectral Modularity algorithm out performs others when applied to the human PostSynaptic Density (PSD) network
Summary
The detection and analysis of community structure in networks has received considerable attention in recent years [1,2,3]. Community structure based clustering is the division of a network into groups of nodes with relatively dense connections within the group and sparser connections to other groups in the network. In many networks, finding the underlying groups provides practical and important information. Groups within social networks for example, may correspond to social units; in biological networks it helps to reveal substructures with common functionality or association with diseases [4,5,6]. Many algorithms exist to solve the problem of breaking up the network into its communities, but most struggle to scale with current biological datasets
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
