epl: 'epl' ExplodeLayout Algorithm.
In UTMB-DIVA-Lab/epl: ExplodeLayout algorithm for network visulization

This package provides functions for the ExplodeLayout (EL) algorithm, described in the following publication: Bhavnani S.K., Chen, T., Ayyaswamy, A., Visweswaran, S., Bellala, G. Enabling Comprehension of Patient Subgroups and Characteristics in Large Bipartite Networks: Implications for Precision Medicine. Proceedings of AMIA Summit on Translational Bioinformatics (2017).

Motivation: Despite strong and significant clustering, many networks look like "hairballs" when visualized using standard layout algorithms such as Kamada Kawai and Fruchterman Rheingold. The ExplodeLayout algorithm addresses this problem by separating given clusters using the analogy of an explode layout drawing utilized to understand complex assemblies in mechanical engineering.

Overview of the EL Algorithm: The EL algorithm takes as input (1) the network data (unipartite or bipartite), (2) node cluster membership (generated from a cluster algorithm such as modularity, or hierarchical clustering), and (3) node layout coordinates (generated from a layout algorithm such as Kamada-Kawai and Fruchterman-Reingold). The algorithm uses the above input to calculate the centroid of each cluster, and moves all nodes in a cluster radially outward such that the centroid is incident to a given circle, and the distances of the nodes to their respective centroid are preserved. Because this movement alters the rotation of the clusters relative to the original layout, the clusters are rotated to match their original orientation. The algorithm generates the radius of the circle through a search, which optimizes the ratio of the overlap of the bounding boxes defining each cluster, to the bounding box defining the entire network. The algorithm uses this optimal explode layout radius to generate a range of layouts in its close vicinity, and displays the network by connecting the nodes based on the inputted network. The EL algorithm has a computational complexity of n.

The algorithm provides parameters that enable the user to (1) input data without layout coordinates and to instruct EL to generate those coordinates using a standard layout algorithm, (2) modify default methods for generating the centroid, exploded coordinates, and range of circle radiuses, and (3) modify how the network is visualized.

#The EL package provides three functions: (1) 'dataConvert' converts user's network data into a format required by EL, (2) 'search' generates a range of exploded network layouts, and (3) 'visualize' generates visualizations of the exploded network layouts.

A sample dataset in the EL input format can be obtained from https://github.com/UTMB-DIVA-Lab/epl) Full documentation can be obtained from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5543384/

UTMB-DIVA-Lab/epl documentation built on July 28, 2019, 5:53 a.m.