perturbCorr: Perturb networks and evaluate subgroup structures of...

In GatesLab/evalClust: Random Perturbation of Count Matrices

Description

This function implements the Masuda, Kojaku & Sano (2018) "Configuration model for correlation matrices preserving the node strength" algorithm for generating correlation matrices that have the same strength distribution as the original matrix. Returns modularity of each randomized correlation matrix.

Usage

perturbCorr(sym.matrix, plot = TRUE, sampleSize = NULL, n = NULL,
  tol = 0.001, stepSize = 0.001, verbose = FALSE)

Arguments

sym.matrix	A symmetric, correlation or covariance matrix object
plot	Logical, defaults to TRUE
sampleSize	The sample size for a correlation matrix, lower values for more sampling error
n	Number of random correlations matrices to return
tol	Tolerance for configuration model convergence. Default to .0001
stepSize	Step size for configuration model NR solver. Increase to increase convergence speed. Default to .001
verbose	Logical, Print convergence information to screen. Defaults to FALSE.

Examples

perturbCorr(examplecorr, sampleSize=100, n=25, plot=FALSE)

GatesLab/evalClust documentation built on Nov. 17, 2019, 3:59 a.m.