clusterExperiment icon indicating copy to clipboard operation
clusterExperiment copied to clipboard

calling RSEC() function error: Error in (function (classes, fdef, mtable) : unable to find an inherited method for function ‘primaryCluster’ for signature ‘"NULL"’

Open siyao-liu opened this issue 6 years ago • 5 comments

Hello,

I'm trying to run RSEC() on a data matrix that contains 1000 genes and 12170 cells. I got the error as below: Error in (function (classes, fdef, mtable) : unable to find an inherited method for function ‘primaryCluster’ for signature ‘"NULL"’

I also found a previous post on github related to this issue: https://github.com/epurdom/clusterExperiment/issues/182

However, the issue was with the clusterMany() function. I wonder if the issue with the RSEC() function has also something to do with the memory? I tried adding the argument "largeDataset=TRUE" in the RSEC() function but it didn't solve the problem. Can anyone help? Thanks very much!

siyao-liu avatar Jan 19 '19 23:01 siyao-liu

Can you provide the full code and the traceback of the error?

epurdom avatar Feb 05 '19 17:02 epurdom

The full code is:

RSEC(seObj, k0s = 3: 15, alphas = c(0.1), betas = 0.8, reduceMethod = "none", clusterFunction = "hierarchical01", minSizes = 1, ncores = NCORES, isCount = FALSE, dendroReduce = "none", dendroNDims=NA, subsampleArgs = list(resamp.num = 100, clusterFunction = "kmeans", clusterArgs = list(nstart = 10)), verbose = TRUE, consensusProportion = 0.7, mergeMethod = "none", random.seed = 424242, consensusMinSize = 10)

The traceback of the error is: Error in (function (classes, fdef, mtable) : unable to find an inherited method for function ‘primaryCluster’ for signature ‘"NULL"’

siyao-liu avatar Feb 05 '19 17:02 siyao-liu

It could be related to memory -- the traceback is not very informative. What machine are you running it on -- in particular what is the memory capacity of the machine (and what is NCORES)?

The original dataset is not very large, but during subsampling, a 12K x 12K matrix is created to store the results, so that's roughly 1.5Gb for each instance. So if you are parallelizing across cores that could add up.

epurdom avatar Feb 20 '19 23:02 epurdom

I'm running it on a cluster. I requested 16 cores, and each has 10.67GB memory capacity.

siyao-liu avatar Feb 22 '19 03:02 siyao-liu

I got the same error and its due to processes being killed by the out-of-mamory handler.

flde avatar Apr 04 '23 09:04 flde