This R repo is a development branch, the actively developed repo is in Python at

Multiscale Graph Correlation (MGC)

CRAN Status Badge eLife Shield Downloads badge Zenodo Shield Travis



In modern scientific discovery, it is becoming increasingly critical to uncover whether one property of a dataset is related to another. The MGC (pronounced magic), or Multiscale Graph Correlation, provides a framework for investigation into the relationships between properties of a dataset and the underlying geometries of the relationships, all while requiring sample sizes feasible in real data scenarios.

Repo Contents

System Requirements

Hardware Requirements

The MGC package requires only a standard computer with enough RAM to support the operations defined by a user. For minimal performance, this will be a computer with about 2 GB of RAM. For optimal performance, we recommend a computer with the following specs:

RAM: 16+ GB CPU: 4+ cores, 3.3+ GHz/core

The runtimes below are generated using a computer with the recommended specs (16 GB RAM, 4 [email protected] GHz) and internet of speed 25 Mbps.

Software Requirements

OS Requirements

This package is supported for Linux operating systems. The package has been tested on the following systems:

Linux: Ubuntu 16.04 Mac OSX: Windows:

Before setting up the MGC package, users should have R version 3.4.0 or higher, and several packages set up from CRAN.

Installing R version 3.4.2 on Ubuntu 16.04

the latest version of R can be installed by adding the latest repository to apt:

sudo echo "deb xenial/" | sudo tee -a /etc/apt/sources.list
gpg --keyserver --recv-key E084DAB9
gpg -a --export E084DAB9 | sudo apt-key add -
sudo apt-get update
sudo apt-get install r-base r-base-dev

which should install in about 20 seconds.

Package dependencies

Users should install the following packages prior to installing mgc, from an R terminal:

install.packages(c('ggplot2', 'reshape2', 'Rmisc', 'devtools', 'testthat', 'knitr', 'rmarkdown', 'latex2exp', 'MASS'))

which will install in about 80 seconds on a recommended machine.

Package Versions

The mgc package functions with all packages in their latest versions as they appear on CRAN on October 15, 2017. Users can check CRAN snapshot for details. The versions of software are, specifically:

ggplot2: 2.2.1
reshape2: 1.4.2
Rmisc: 1.5
devtools: 1.13.3
testthat: 0.2.0
knitr: 1.17
rmarkdown: 1.6
latex2exp: 0.4.0
MASS: 7.3

If you are having an issue that you believe to be tied to software versioning issues, please drop us an Issue.

Installation Guide

From an R session, type:

install_github('neurodata/r-mgc', build_vignettes=TRUE)  # install mgc with the vignettes
require(mgc)  # source the package now that it is set up
vignette("MGC", package="mgc")  # view one of the basic vignettes

The package should take approximately 20 seconds to install with vignettes on a recommended computer.

Instructions for Use

Please see the vignettes for help using the package:

vignette("MGC", package="mgc")
vignette("Discriminability", package="mgc")
vignette("simulations", package="mgc")


Pseudocode for the methods employed in the mgc package can be found on the arXiv - MGC in Appendix C (starting on page 30).


For citing code or the paper, please use the citations found in citation.bib.



All the code to reproduce any figures from is available here


Here, we describe how to reproduce the manuscript figures from the discriminability paper. To begin, clone this repository locally:

git clone

We assume that the directory r-mgc placed locally on the system is <package_root>. Note that all figures were stylized using Adobe Photoshop prior to submission.


Note: the scripts will automatically multithread, however, the simulation benchmarks take quite a while to execute (1.5 days on a 96 core machine with 1 TB of RAM).

Using the included bound, one sample, and two sample data, you can proceed to duplicate the figure by opening the R notebook simulation plots, and executing the script.

# edit lines 17 and 18, and lines 210 and 211, and set to your local path where
# preprocessed brains are located
source('./real_data_driver.R')  # runs the discriminability calculations
# edit lines 17 and 18, and lines 108 and 109, to the location of the 
# preprocessed brains
source('./realdat_perm_testing.R')  # runs the two sample testing

Again, the scripts will multithread, but can be expected to take approximately 3 days on a 96 core, 1 TB RAM machine.

To regenerate Figure 2 from the manuscript, users can execute the 64 Pipelines Figure notebook.


Results can be expected to take 2 days on a 96 core, 1 TB machine.

To reproduce the figure, users can use the Effect Size Investigation notebook.

neurodata/r-mgc documentation built on Dec. 25, 2019, 4:13 p.m.