Error: failed to stop 'SOCKcluster' cluster: error writing to connection

[ttriche]

This is a new development; I'm not sure what's behind it. But it's killing every dmrseq run now...

Retaining 561544 loci.
Assuming the test covariate tumor is a factor.
Condition: 1 vs 0
Parallelizing using 40 workers/cores (backend: BiocParallel:MulticoreParam).

Detecting candidate regions with coefficient larger than 0.1 in magnitude.
...Chromosome chr22: Error in serialize(data, node$con, xdr = FALSE) : 
  error writing to connection
Error: failed to stop 'SOCKcluster' cluster: error writing to connection

Running the -3.4 branch from github as of 4/23/2018:

R> sessionInfo()
R version 3.4.3 (2017-11-30)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: CentOS Linux 7 (Core)

Matrix products: default
BLAS: /primary/projects/triche/lib64/R/lib/libRblas.so
LAPACK: /primary/projects/triche/lib64/R/lib/libRlapack.so

locale:
[1] C

attached base packages:
[1] stats4    parallel  stats     graphics  grDevices datasets  utils
[8] methods   base

other attached packages:
 [1] biscuitEater_0.9.11        bsseq_1.14.0
 [3] SummarizedExperiment_1.8.1 DelayedArray_0.4.1
 [5] matrixStats_0.53.1         Biobase_2.38.0
 [7] GenomicRanges_1.30.3       GenomeInfoDb_1.14.0
 [9] IRanges_2.12.0             S4Vectors_0.16.0
[11] BiocGenerics_0.24.0        BiocInstaller_1.28.0
[13] skeletor_1.0.4             magrittr_1.5
[15] gtools_3.5.0               useful_1.2.3
[17] ggplot2_2.2.1              purrr_0.2.4
[19] knitr_1.20

loaded via a namespace (and not attached):
  [1] colorspace_1.3-2
  [2] XVector_0.18.0
  [3] roxygen2_6.0.1
  [4] bit64_0.9-7
  [5] interactiveDisplayBase_1.16.0
  [6] AnnotationDbi_1.40.0
  [7] qualV_0.3-3
  [8] xml2_1.2.0
  [9] codetools_0.2-15
 [10] R.methodsS3_1.7.1
 [11] impute_1.52.0
 [12] dmrseq_0.99.10

[kdkorthauer]

Hi @ttriche,

Thanks for the bug report. I think it is related to these recent issues in bsseq: hansenlab/bsseq#65 and DelayedArray: Bioconductor/DelayedArray#16

I'll follow up with the bsseq/DelayedArray authors about a fix.

Best, Keegan

[kdkorthauer]

Hi @ttriche,

The issue has been fixed by the bsseq authors by swapping dependency on MatrixStats for DelayedMatrixStats. I can make this change for dmrseq if you are still getting the error after installing the newest version of bsseq from hansenlab/bsseq. Let me know; I have been unable to reproduce the error on my machines.

Best, Keegan

[kdkorthauer]

Hi @ttriche,

I have updated the package to use the rowSums2, rowMeans2, colSums2 and colMeans2 functions from DelayedMatrixStats. I hope this resolves the issues you've been experiencing. These changes should propagate to the release version of Bioc 3.7 on May 1; in the meantime it's safer to install from Github during the transition.

Best, Keegan

[ttriche]

Right on -- will do. Thanks for the fixes -- still playing with the various options for dmrseq to tune it for [e]RRBS and mid-coverage WGBS -- but it's a great framework. My undergraduate intern who's planning to work on N-of-1 permutations starts on May 14th, so it's great to resolve this now rather than then :-)

Much obliged.

--t

On Sun, Apr 29, 2018 at 5:13 PM, Keegan Korthauer notifications@github.com wrote:

Hi @ttriche https://github.com/ttriche,

I have updated the package to use the rowSums2, rowMeans2, colSums2 and colMeans2 functions from DelayedMatrixStats. I hope this resolves the issues you've been experiencing. These changes should propagate to the release version of Bioc 3.7 on May 1; in the meantime it's safer to install from Github during the transition.

Best, Keegan

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/kdkorthauer/dmrseq/issues/10#issuecomment-385282209, or mute the thread https://github.com/notifications/unsubscribe-auth/AAARIkC6pxp3WI-MjcphLEf2zfy0yrZRks5ttizfgaJpZM4Thmrb .

[ttriche]

this is fixed by the rowMeans2/rowSums2 updates. Thanks Keegan!