100 A Practical Guide to Cancer Systems Biology
up.4.fold <- samrSummTable$genes.up[,2][as.numeric(samrSummTable
$genes.up[,7])>=4]
dn.4.fold <- samrSummTable$genes.lo[,2][as.numeric(samrSummTable
$genes.lo[,7])<=1/4]
The compCorrGraph() takes an ExpressionSet object as the argument and
the parameter tau is the cutoff beyond which absolute PCC values will
be plotted in the graph as visible undirected edges. Here, you can plot a
correlation graph in ALL1/AF4 samples (N=10) for these DE genes under
|PCC|> 0 .7 (corresponding to a two-tailedP-value = 0.024; Fig. 2):
eMat<- data.all1Af4[rownames(data.all1Af4) %in% c(up.4.fold, dn.4.
fold),]
rownames(eMat)<- sapply(strsplit(rownames(eMat), “\/”), “[”)[1,]
eSet<- new(“ExpressionSet”, exprs=eMat)
corrG<- compCorrGraph(eSet, tau=0.7)
edgemode(corrG)<-“undirected”
plot(corrG)
Figure 2. Gene correlation network for B cell leukemia with ALL1−AF4 rearrangement.