correlation matrix to build networks

Question

I have used the MixOmics package in R for a two matrices (canonical correlation analysis) and I have a resultant correlation matrix. I would like to build a correlation network from the result obtained. I earlier thought of using the gene set correlation analysis package but I do not know how to install it and there are no sources over the internet to install it in R (http://www.biostat.wisc.edu/~kendzior/GSCA/).

Also could you suggest what other packages I could use to build networks with correlation matrix as input ?? I thought of Rgraphviz but do not know if it is possible.

Answer 1

Copying this answer mostly from my previous answer at https://stackoverflow.com/a/7600901/567015

The qgraph package is mostly intended to visualize correlation matrices as a network. This will plot variables as nodes and correlations as edges connecting the nodes. Green edges indicate positive correlations and red edges indicate negative correlations. The wider and more saturated the edges the stronger the absolute correlation.

For example (this is the first example from the help page), the following code will plot the correlation matrix of a 240 variable dataset.

library("qgraph")
data(big5)
data(big5groups)
qgraph(cor(big5),minimum=0.25,cut=0.4,vsize=2,groups=big5groups,legend=TRUE,borders=FALSE)
title("Big 5 correlations",line=-2,cex.main=2)

You can also cluster strongly correlated nodes together (uses Fruchterman-Reingold) which creates quite a clear image of what the structure of your correlation matrix actually looks like:

For an extensive introduction take a look at http://www.jstatsoft.org/v48/i04/paper