corr_comm_detection

Multilayer community detection with covariance matrix input

corr_comm_detection.py

Contains the following functions

• cov_to_corr: function to transform a covariance matrix into a correlation matrix

• con_corr_func: function to generate a configuration model correlation matrix from an empirical correlation or covariance matrix, using configcorr package

• multicorrcat: function (inspired by multicat.m) to output a flattened modularity matrix

• it_genlouvain_corr_consensus: function to run iterated GenLouvain, a community detection algorithm, on the flattened modularity matrix

• corr_partition_info: function to obtain information about the partition needed for significance calculations

• corr_intra_z: function to calculate significance (Z score for total intralayer weight) of each community in the partition

• main: main function that uses all the above functions to perform community detection with covariance matrix/matrices input and outputs the partition and the Z score for each community detected

Python package dependencies:

corr_comm_detection has the following dependencies. These packages need to be installed manually.

• numpy
• pandas
• matplotlib
• scipy
• configcorr
• netneurotools
• matlab.engine
• Also must download the GenLouvain matlab package

multilayer_example.ipynb

Contains example usage of corr_comm_detection with dummy data.

partition_corr_gamma3.xlsx

Contains a table of the partition obtained by corr_comm_detection with resolution parameter gamma=3. Each row corresponds to a gene, each column corresponds to a tissue (i.e., a layer of the network), and each entry is the index of the community to which the corresponding node belongs.