YosefLab / scone

53 stars 12 forks source link

bplapply memory issue when running scone(run=TRUE) #69

Open ckmah opened 8 years ago

ckmah commented 8 years ago

I came across an error running the following call to scone(run=TRUE) while running through the Bioc2016 workshop vignette (added verbose = TRUE flag):

res <- scone(expr = as.matrix(fcounts),scaling = c(none = identity, deseq = DESEQ_FN, tmm = TMM_FN, uqp = UQ_FN_POS, fq = FQT_FN),
+              ruv_negcon = fhk, k_ruv = 3,
+              qc = as.matrix(fqc), k_qc = 3,
+              bio = fbio,adjust_bio = "yes",
+              batch = fbatch,adjust_batch = "yes",
+              run = TRUE,params = params,
+              eval_poscon = fde, eval_kclust = 2:3, verbose = TRUE)

Output:

Imputation step...
Scaling step...
Computing RUV factors...
Error: 'bplapply' receive data failed:
  error reading from connection

Some details: fcounts is a 10610 x 91 expression matrix using the data from the vignette. I ran this through RStudio with R default memory settings on OSX with 16GB RAM. @mbcole advised me to post this issue as a potential memory bug.

mbcole commented 8 years ago

Hi Clarence,

Thanks for posting this error. I'm actually quite surprised this is occurring for so few cells (91, correct?). I run SCONE on larger data sets using a laptop with less memory. I'm curious whether the error occurs in serial mode. This may be specified by calling

`BiocParallel::register(BiocParallel::SerialParam())`

prior to the scone() call.

ckmah commented 8 years ago

I just tried it in serial mode and the error seems to have gone away. I'll keep you updated and let you know when I get a chance to try it on a larger dataset.

kieran-mace commented 6 years ago

I have the same issue when I run DESeq2 with the parallel option