Support‎ > ‎

Graph Theory

We are pleased to announce a new Methods Core release: Graph Theory. It is now in your lab's Methods Core repository. 

Graph theory is a newer method that quantifies high-level patterns in connectivity. The brain is first parcellated into a large number of nodes, a connectivity value is calculated for each pair of nodes, and then the resulting "connectome" is submitted for graph theoretic analysis. For a great summary of the rationale for, and limitations of, graph theory see the short article from Steve Taylor from the most recent Biological Psychiatry.

Our graph theory tool allows you to calculate a large number of graph metrics including: Degree, Clustering, Characteristic Path Length, Transitivity, Global Efficiency, Modularity, Assortativity, Small Worldness, among others.
This user-friendly script has many nice features:
  • lets you calculate these metrics on the whole brain or on certain networks of interest. 
  • produces "first-level" outputs as well as "second-level" outputs stored in easy to understand CSVs. 
  • calculates second-level outputs using t-tests or non-parametric permutation tests (often demanded by reviewers). 
  • also calculates certain metrics, e.g., degree, on a "per node" basis, and writes out second-level 3d maps.
The Graph Theory script was developed primarily by Yu Fang (a member of Chandra's lab), with help from Mike Angstadt and Chandra Sripada. Because of the complexity of the method, we do not attempt to provide step-by-step documentation. If using the script for a project, we encourage you to involve at least some members of the development team as collaborators. 

References