Distance Backbone Toolbox

[robertjosephpalermo]

New plugin or plugin update?

• New Plugin
• Update

What is the purpose of this plugin?

This plugin allows you to compute the metric and ultrametric distance backbones defined in Simas et al 2021 [https://doi.org/10.1093/comnet/cnab021]. For larger networks, we recommend using the Python library distanceclosure [https://github.com/CASCI-lab/distanceclosure].

Those backbones are a parameter-free and algebraically-principles network sparsification methodologies which remove edges that break a generalized triangular inequality and, as a consequence, are redundant for shortest-path computations. Interestingly, the ultrametric backbone expands the concept of a minimum spanning tree to directed graphs and is itself a subgraph of the metric backbone for both directed and undirected networks.

How to test your plugin in Gephi?

1. Load a weighted graph (preferably graphml format) into Gephi, in which there is a variable labeled as "weight" that represents a notion of distance (smaller values are stronger associations).

2. Load the plug in, and this will create two buttons under the statistics tab: Metric Backbone, and Ultrametric Backbone. To test functionality, first press 'run' on the Metric Backbone button to compute the metric backbone, then press 'run' in the Ultrametric Backbone button to compute the ultrametric backbone.

3. Note: since the ultrametric backbone is a subgraph of the metric backbone you should compute the metric and then the ultrametric to test both. However, the opposite will lead to wrong results (the graph will not change).

Checklist before submission

• Did you merged with the master branch to get the latest updates?
• Did you build and test the plugin successfully locally?
• Did you include metadata (e.g. plugin author, license) in the pom.xml file
• Did you make sure NOT to include any unecessary files in your PR?
• Did you write unit tests to test your plugin functionalities?