Hi.

I installed yesterday, on two ubuntu machines, BHC by

source("http://bioconductor.org/biocLite.R")
biocLite("BHC")

I tried:

RNAdata=read.table("http://bass.uib.es/~jairo/data/Datatest")

require(BHC)

hc1 <- bhc(RNAdata,dimnames(RNAdata)[[1]],0,verbose=TRUE)

but it aborts. Originally, I tried without the 0 in the third parameter but it also breaks down.

Thanks for any hint,

Jairo

University of the Balearic Islands

Loading required package: BHC
[1] Running Bayesian Hierarchical Clustering....
[1] "DataType: multinomial"
[1] Optimising global hyperparameter...

 *** caught segfault ***
address 0x28, cause 'memory not mapped'

Traceback:
 1: .C("bhcWrapper_multinomial", as.integer(data), as.integer(nDataItems),     as.integer(nFeatures), as.double(globalHyperParameter), as.integer(nFeatureValues),     logEvidence = as.double(123), node1 = vector(mode = "integer",         length = nDataItems - 1), node2 = vector(mode = "integer",         length = nDataItems - 1), mergeOrder = vector(mode = "integer",         length = nDataItems - 1), mergeWeight = vector(mode = "numeric",         length = nDataItems - 1))
 2: RunBhcWrapper(param.lower, dataTypeID, data, timePoints, nDataItems,     nFeatures, nFeatureValues, noise = noise)
 3: FindOptimalHyperparameter(dataTypeID, data, timePoints, noise,     nDataItems, nFeatures, nFeatureValues, verbose = verbose)
 4: bhc(RNAdata, dimnames(RNAdata)[[1]], verbose = TRUE)
 5: eval(expr, envir, enclos)
 6: eval(ei, envir)
 7: withVisible(eval(ei, envir))
 8: source("BHCtest.R")

No sessionInfo() output as R crashes.

Hi Jairo,

I believe the reason for the crash is that you're giving BHC real-valued data for the default data model (which is multinomial, so requires discrete values).  The documentation for the 'bhc' function gives an example of how to discretise your data.  Adapting that, you could try the following:

require(BHC)

RNAdata     = read.table("http://bass.uib.es/~jairo/data/Datatest")
RNAdata    = t(RNAdata)
itemLabels = dimnames(RNAdata)[[2]]

percentiles   = FindOptimalBinning(RNAdata, itemLabels, transposeData=TRUE, verbose=TRUE)
discreteData = DiscretiseData(RNAdata, percentiles=percentiles)
hc1               = bhc(t(discreteData), itemLabels, verbose=TRUE)
plot(hc1)

Best regards,

Rich