SCANNET: A Software for Identification of Communities in Networks

Abstract

The Network Similarity Analysis (NSA) method is a powerful tool for identifying communities in weighted networks whose links depend on the edge's weight. The method can be applied to different types of networks, with nodes and edges receiving different interpretations. For instance, the network can be constructed based on the similarity between proteins that play the same function in distinct species, or between DNA sequences that share the same evolutionary origins. The original implementation of the NSA method consisted of pipeline of algorithms implemented in different programming languages, making it difficult to use. SCANNET is a software developed to integrate all algorithms of the NSA method into a unique computational environment, to perform all the steps needed to identify communities in networks in a single workflow. As an example for this paper, we applied SCANNET to implement a single workflow that correctly identified the communities in protein networks containing different sequences of enzymes. SCANNET has correctly generated the protein networks, the dendrograms, and the color representation of the neighborhood matrix for each of the studied case. The experiments have shown that SCANNET is a simple, efficient, reliable, and intuitive implementation of the NSA method for identifies communities in the case of protein networks. In addition, SCANNET generates charts with all the distances between pairs of neighboring networks, which facilitate the identification of critical networks for community identification. SCANNET also generates dendrograms, and color representations of neighborhood matrices that can be manipulated by different analytical software, such as Origin Lab or Matlab, offering a friendly interface.