colorGraph {SEMgraph}

Vertex and edge graph coloring on the base of fitting

Description

Add vertex and edge color attributes to an igraph object, based on a fitting results data.frame generated by SEMrun.

Usage

colorGraph(
  est,
  graph,
  group,
  method = "none",
  alpha = 0.05,
  vcolor = c("lightblue", "white", "pink"),
  ecolor = c("royalblue3", "gray50", "red2"),
  ewidth = c(1, 2),
  ...
)

Arguments

est

A data.frame of estimated parameters and p-values, derived from the fit object returned by SEMrun. As an alternative, the user may provide a "gest" or "dest" data.frame generated by SEMrun.

graph

An igraph object.

group

group A binary vector. This vector must be as long as the number of subjects. Each vector element must be 1 for cases and 0 for control subjects.

method

Multiple testing correction method. One of the values available in p.adjust. By default, method is set to "none" (i.e., no multiple test correction).

alpha

Significance level for node and edge coloring (by default, alpha = 0.05).

vcolor

A vector of three color names. The first color is given to nodes with P-value < alpha and beta < 0, the third color is given to nodes with P-value < alpha and beta > 0, and the second is given to nodes with P-value > alpha. By default, vcolor = c("lightblue", "white", "pink").

ecolor

A vector of three color names. The first color is given to edges with P-value < alpha and regression coefficient < 0, the third color is given to edges with P-value < alpha and regression coefficient > 0, and the second is given to edges with P-value > alpha. By default, vcolor = c("blue", "gray50", "red2").

ewidth

A vector of two values. The first value refers to the basic edge width (i.e., edges with P-value > alpha), while the second is given to edges with P-value < alpha. By default ewidth = c(1, 2).

...

Currently ignored.

Value

An igraph object with vertex and edge color and width attributes.

Author(s)

Mario Grassi [e-mail]

Examples

# Model fitting: node perturbation
sem1 <- SEMrun(graph = alsData$graph, data = alsData$exprs,
               group = alsData$group,
               fit = 1)
est1 <- parameterEstimates(sem1$fit)

# Model fitting: edge perturbation
sem2 <- SEMrun(graph = alsData$graph, data = alsData$exprs,
               group = alsData$group,
               fit = 2)
est20 <- subset(parameterEstimates(sem2$fit), group == 1)[, -c(4, 5)]
est21 <- subset(parameterEstimates(sem2$fit), group == 2)[, -c(4, 5)]

# Graphs
g <- alsData$graph
x <- alsData$group

old.par <- par(no.readonly = TRUE)
par(mfrow=c(2,2), mar=rep(1,4))
gplot(colorGraph(est = est1, g, group = x, method = "BH"),
      main = "vertex differences")
gplot(colorGraph(est = sem2$dest, g, group = NULL),
      main = "edge differences")
gplot(colorGraph(est = est20, g, group = NULL),
      main = "edges for group = 0")
gplot(colorGraph(est = est21, g, group = NULL),
      main = "edges for group = 1")
par(old.par)