showNetwork: A function for looking at the co-expression among a small group of genes

Description

This function draws a network for a selected group of genes using igraph. The edges are colored in accordance with the correlation strength indicated by the inputted D matrix, ranging from red (strong negative correlation) to blue (strong positive correlation)

Usage

showNetwork(geneSet, D, condFocus, gsep = "~", layout = "kamada.kawai", seed = NULL, hidingThreshold=NULL, ...)

Arguments

geneSet

An array of genes of interest; should not be larger than a dozen or so

The correlation matrix output of makeMyD()

condFocus

The condition of interest for this network. Should be one of the integers in the conditions array

gsep

A separator that indicates a gene-pair, such as P53~MAPK1. The separator should not appear in any of the gene names

layout

A layout to be parsed and used by igraph. Examples include circle (the default) and kamada.kawai; see the documentation for igraph for more information. At this time it is not possible to specify parameters specific to particular layouts

seed

A seed to be set before invoking igraph's layout generation. This is useful for layouts such as random, where node postion is not deterministic

hidingThreshold

A threshold which we will shorthand by 'h'. If this value is non-NULL, all correlations in [-h, h] will not be plotted in the network. This is useful for removing clutter in busy networks will relatively high (say, 20+) numbers of genes

...

Other options to be passed to plot.igraph(). Networks generated by igraph require quite a bit of formatting, and it is up to the user to do so by specifying appropriate options from the following:

vertex.shape=, vertex.label.cex=, vertex.color=, vertex.frame.color=, vertex.size=, vertex.label.color=, vertex.label.family=, and edge.width=

The following options are hard-coded and may not be overwritten:

vertex.label=geneSet, edge.arrow.mode=0, edge.color=[red/blue colors]

where [red/blue colors] is determined by the correlation information contained in D, possibly overwritten in some cases if hidingThreshold is non-NULL

Value

References

Dawson JA and Kendziorski C. An empirical Bayesian approach for identifying differential co-expression in high-throughput experiments. (2011) Biometrics. E-publication before print: http://onlinelibrary.wiley.com/doi/10.1111/j.1541-0420.2011.01688.x/abstract

Examples

Run this code

data(fiftyGenes)
tinyCond <- c(rep(1,100),rep(2,25))
tinyPat <- ebPatterns(c("1,1","1,2"))
D <- makeMyD(fiftyGenes, tinyCond, useBWMC=TRUE)
twentyGeneNames <- dimnames(fiftyGenes)[[1]][c(1:10,26:35)]

showNetwork(twentyGeneNames, D, condFocus = 1, gsep = "~",
  layout = "kamada.kawai", seed = 5, vertex.shape="circle",
  vertex.label.cex=1, vertex.color="white", edge.width=2,
  vertex.frame.color="black", vertex.size=20,
  vertex.label.color="black", vertex.label.family="sans",
  hidingThreshold=0.3)
#
showNetwork(twentyGeneNames, D, condFocus = 2, gsep = "~",
  layout = "kamada.kawai", seed = 5, vertex.shape="circle",
  vertex.label.cex=1, vertex.color="white", edge.width=2,
  vertex.frame.color="black", vertex.size=20,
  vertex.label.color="black", vertex.label.family="sans",
  hidingThreshold=0.3)
#