Good morning, I succesfully created a network using the following based on an Affy GCRMA normalized eset:

net = blockwiseModules(datExpr, power = 4,
                       TOMType = "unsigned", minModuleSize = 30,
                       reassignThreshold = 0, mergeCutHeight = 0.25,
                       numericLabels = TRUE, pamRespectsDendro = FALSE,
                       saveTOMs = TRUE,
                       saveTOMFileBase = "2-GSE10325_CD19B_TOM",
                       verbose = 3,
                       maxBlockSize = 20000)

Unfortunately the TOM object created can't be subset to a module for Cytoscape export:

dissTOM = 1-TOM

# Select modules
module = c("turquoise");

# Select module probes
probes = colnames(datExpr)
inModule = (moduleColors==module);
modProbes = probes[inModule]; # 3505 probes in module turquoise

# Recalculate intramodular connectivity (IMConn)
IMConn = softConnectivity(datExpr[, modProbes]);

# Select the corresponding Topological Overlap
length(inModule)
modTOM = dissTOM[inModule, inModule];
dimnames(modTOM) = list(modProbes, modProbes)

Error in dissTOM[inModule, inModule] : incorrect number of dimensions

I'm able to interrogate dissTOM using min(dissTOM) and max(dissTOM), and obviously this TOM functioned so far as to provide the basis for network creation and ensuing module creation. I used the same Affy GCRMA normalized eset to generate a network manually:

softPower = 14;
adjacency_sp14_bicor_signed = adjacency(datExpr, power= softPower, type="signed", corFnc="bicor")
TOM_sp14_bicor_signed = TOMsimilarity(adjacency_sp14_bicor_signed, TOMType="signed", verbose=5)
dissTOM_sp14_bicor_signed = 1-TOM_sp14_bicor_signed
etc etc

This produced a TOM which I was able to subset to Cytoscape beautifully, but unfortunately I can't use this network due to historical purposes (I have earlier results based on the unsigned network which I'm moving forward with for now).

Is there something about the automatic network creation (blockwiseModules) that is creating a TOM which can't be subset? How can I repair or manipulate my current TOM?

Thanks,

Robert D. Robl

TOM is saved as data from a distance structure; I realized this is not spelled anywhere in the help files, so I'll add the information. When you load it, issue TOM.mat = as.matrix(TOM) and work with TOM.mat.

Note that in the tutorial code, modTOM is assumed to be a similarity, not dissimilarity (which is what you use). You may want to keep this distinction in mind.

HTH,

Peter