nethet (version 1.4.2)

dot_plot: Create a plot showing the edges with the highest partial correlation in any cluster.

Description

This function takes the output of het_cv_glasso or mixglasso and creates a plot of the highest scoring edges along the y axis, where, the edge in each cluster is represented by a circle whose area is proportional to the smallest mean of the two nodes that make up the edge, and the position along the y axis shows the partial correlation of the edge.

Usage

dot_plot(net.clustering, p.corrs.thresh = 0.25, hard.limit = 50,
  display = TRUE, node.names = rownames(net.clustering$Mu),
  group.names = sort(unique(net.clustering$comp)), dot.size.range = c(3,
  12))

Arguments

net.clustering

A network clustering object as returned by het_cv_glasso or mixglasso.

p.corrs.thresh

Cutoff for the partial correlations; only edges with absolute partial correlation > p.corrs.thresh (in any cluster) will be displayed.

hard.limit

Additional hard limit on the number of edges to display. If p.corrs.thresh results in more edges than hard.limit, only hard.limit edges with the highest partial correlation are returned.

display

If TRUE, print the plot to the current output device.

node.names

Names for the nodes in the network.

group.names

Names for the clusters or groups.

dot.size.range

Graphical parameter for scaling the size of the circles (dots) representing an edge in each cluster.

Value

• Returns a ggplot2 object. If display=TRUE, additionally displays the plot.

Examples

n = 500
p = 10
s = 0.9
n.comp = 3

# Create different mean vectors
Mu = matrix(0,p,n.comp)

# Define non-zero means in each group (non-overlapping)
nonzero.mean = split(sample(1:p),rep(1:n.comp,length=p))

# Set non-zero means to fixed value
for(k in 1:n.comp){
	Mu[nonzero.mean[[k]],k] = -2/sqrt(ceiling(p/n.comp))
}

# Generate data
sim.result = sim_mix_networks(n, p, n.comp, s, Mu=Mu)
mixglasso.result = mixglasso(sim.result$data, n.comp=3)
mixglasso.clustering = mixglasso.result$models[[mixglasso.result$bic.opt]]

dot_plot(mixglasso.clustering, p.corrs.thresh=0.5)