A new grid- and modularity-based layout algorithm for complex biological networks.

Sheng He,Yi-Jun Liu,Fei-Yue Ye,Ren-Jun Dai,Ren-Pu Li

doi:10.1371/journal.pone.0221620

Sheng He, Yi-Jun Liu + Show 3 more

Open Access

https://doi.org/10.1371/journal.pone.0221620

Copy DOI

Journal: PloS one	Publication Date: Aug 29, 2019
Citations: 4	License type: CC BY 4.0

Affiliation: Jiangsu University of Technology

Abstract

The visualization of biological networks is critically important to aid researchers in understanding complex biological systems and arouses interest in designing efficient layout algorithms to draw biological networks according to their topology structures, especially for those networks with potential modules. The algorithms of grid layout series have an advantage in generating compact layouts with overlap-free nodes compared to force-directed; however, extant grid layout algorithms have difficulty in drawing modular networks and often generate layouts of high visual complexity when applied to networks with dense or clustered connectivity structure. To specifically assist the study of modular networks, we propose a grid- and modularity-based layout algorithm (GML) that consists of three stages: network preprocessing, module layout and grid optimization. The algorithm can draw complex biological networks with or without predefined modules based on the grid layout algorithm. It also outperforms other existing grid-based algorithms in the measurement of computation performance, ratio of edge-edge/node-edge crossings, relative edge lengths, and connectivity F-measures. GML helps users to gain insight into the network global characteristics through module layout, as well as to discern network details with grid optimization. GML has been developed as a VisANT plugin (https://hscz.github.io/Biological-Network-Visualization/) and is freely available to the research community.

Highlights

Network diagrams provide a fundamental conceptual framework for visualizing and mining high-throughput biological datasets, as well as for gaining insights and interpreting the biological implications by means of graph drawing algorithms [1,2,3]
For complex biological networks composed of thousands of nodes, drawing algorithms may strive to grasp the global characteristics of networks to clarify their complexity [2,6,7]
The hybrid grid layout and gridbased layout algorithm (GBL) are modularity -based algorithms that achieve relatively good performance compared with other grid layouts [20,22]

Summary

Introduction

Network diagrams provide a fundamental conceptual framework for visualizing and mining high-throughput biological datasets, as well as for gaining insights and interpreting the biological implications by means of graph drawing algorithms [1,2,3]. For complex biological networks composed of thousands of nodes, drawing algorithms may strive to grasp the global characteristics of networks to clarify their complexity [2,6,7]. Modularization is one of the most significant global characteristics of biological networks, where closely connected nodes (e.g., biomolecules) are usually organized as a module to carry out a specific function [4,8,9]. Biological modules can be generated through clustering algorithms (i.e., pseudomodules) that aim to identify sets of closely connected nodes from.

Methods

Results

Conclusion