Questions related to BSmooth error: encountered errors: 2 of 2 smoothing tasks failed.

[Lxia626]

Hi, I'm running BSmooth using the command BSmooth(bismark, ns=10, h=1000, maxGap=1e5, verbose=T, BPPARAM = MulticoreParam(workers = 32, progressbar = TRUE)), which is on a dataset generated from GpC methylation data and the result of bismark shows like this:
An object of type 'BSseq' with
9424795 methylation loci
1 samples
has not been smoothed
All assays are in-memory.
However, the error said:
Error in BSmooth(bismark, ns = 10, h = 1000, maxGap = 1e+05, verbose = T, :
BSmooth() encountered errors: 2 of 2 smoothing tasks failed.

I have successfully running the same command on a much smaller dataset which contains less them 5,000 GpC methylation sites on 2 samples. I don't know what was wrong when I was running on big dataset.

Here follows my sessionInfo():
R version 4.4.3 (2025-02-28)
Platform: x86_64-conda-linux-gnu
Running under: CentOS Linux 7 (Core)

Matrix products: default
BLAS/LAPACK: /home/path/software/miniconda/envs/PA_env/lib/libopenblasp-r0.3.30.so; LAPACK version 3.12.0

locale:
[1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C
[3] LC_TIME=en_US.UTF-8 LC_COLLATE=en_US.UTF-8
[5] LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8
[7] LC_PAPER=en_US.UTF-8 LC_NAME=C
[9] LC_ADDRESS=C LC_TELEPHONE=C
[11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C

time zone: XXX
tzcode source: system (glibc)

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

other attached packages:
[1] BiocParallel_1.40.0 Rsamtools_2.22.0
[3] Biostrings_2.74.0 XVector_0.46.0
[5] dplyr_1.1.4 tibble_3.3.0
[7] bsseq_1.42.0 SummarizedExperiment_1.36.0
[9] Biobase_2.66.0 MatrixGenerics_1.18.0 [11] matrixStats_1.5.0 GenomicRanges_1.58.0
[13] GenomeInfoDb_1.42.0 IRanges_2.40.0
[15] S4Vectors_0.44.0 BiocGenerics_0.52.0
[17] data.table_1.17.6

loaded via a namespace (and not attached):
[1] rjson_0.2.23 rhdf5_2.50.0
[3] lattice_0.22-7 generics_0.1.4
[5] vctrs_0.6.5 rhdf5filters_1.18.0
[7] tools_4.4.3 bitops_1.0-9
[9] curl_6.2.2 parallel_4.4.3
[11] pkgconfig_2.0.3 R.oo_1.27.1
[13] Matrix_1.7-3 BSgenome_1.74.0
[15] RColorBrewer_1.1-3 sparseMatrixStats_1.18.0
[17] lifecycle_1.0.4 GenomeInfoDbData_1.2.13
[19] compiler_4.4.3 farver_2.1.2
[21] statmod_1.5.0 codetools_0.2-20
[23] permute_0.9-8 RCurl_1.98-1.16
[25] yaml_2.3.10 pillar_1.11.0
[27] crayon_1.5.3 R.utils_2.13.0
[29] DelayedArray_0.32.0 limma_3.62.1
[31] abind_1.4-5 gtools_3.9.5
[33] tidyselect_1.2.1 locfit_1.5-9.12
[35] restfulr_0.0.16 grid_4.4.3
[37] colorspace_2.1-1 cli_3.6.5
[39] SparseArray_1.6.0 magrittr_2.0.3
[41] S4Arrays_1.6.0 XML_3.99-0.17
[43] DelayedMatrixStats_1.28.0 scales_1.4.0
[45] UCSC.utils_1.2.0 httr_1.4.7
[47] R.methodsS3_1.8.2 HDF5Array_1.34.0
[49] BiocIO_1.16.0 rtracklayer_1.66.0
[51] rlang_1.1.6 Rcpp_1.1.0
[53] glue_1.8.0 BiocManager_1.30.26
[55] jsonlite_2.0.0 R6_2.6.1
[57] Rhdf5lib_1.28.0 GenomicAlignments_1.42.0
[59] zlibbioc_1.52.0

Here is BiocManager::valid():
Bioconductor version '3.20'

• 16 packages out-of-date
• 0 packages too new

create a valid installation with

BiocManager::install(c(
"abind", "assorthead", "BiocParallel", "Biostrings", "curl", "data.table",
"DelayedMatrixStats", "GenomeInfoDb", "IRanges", "limma", "MatrixGenerics",
"RCurl", "rhdf5", "rhdf5filters", "SparseArray", "XML"
), update = TRUE, ask = FALSE, force = TRUE)

more details: BiocManager::valid()$too_new, BiocManager::valid()$out_of_date

Warning message:
16 packages out-of-date; 0 packages too new

[PeteHaitch]

I suggest trying with a smaller number of workers, perhaps BPPARAM = MulticoreParam(workers = 8, progressbar = TRUE).
Unfortunately, it's not uncommon for (hard to reproduce) errors to occur when using larger number of workers, particularly on shared systems.

[Lxia626]

Hi, thanks for your reply. Unfortunately, I have tried workers = 8 or 4, and even 2, ALL of them failed with the same error: Error in BSmooth(bismark, ns = 10, h = 100, maxGap = 1e+05, verbose = TRUE, :
BSmooth() encountered errors: 2 of 2 smoothing tasks failed.

Also, there is no detailed information on what the error is, only this sentence, which confused me most and I have no idea about how to debug...

[PeteHaitch]

Re-looking at your original post, your dataset has only 1 sample - is that correct?
I think there won't be much advantage to using multiple workers if there is only 1 sample, so can you try running again with BPPARAM = SerialParam()?

I agree that it can be frustrating to figure out the source of such errors and that the error seemingly being hidden doesn't help.

[kasperdanielhansen]

What sometimes happens is that you have a contig with very, very few CpGs in it and then the smoothing fails. It is hard to know if this is the cause, but it is easy to check. Best, Kasper

…

On Wed, Jul 30, 2025 at 11:50 PM Peter Hickey ***@***.***> wrote: *PeteHaitch* left a comment (hansenlab/bsseq#151) <#151 (comment)> Re-looking at your original post, your dataset has only 1 sample - is that correct? I think there won't be much advantage to using multiple workers if there is only 1 sample, so can you try running again with BPPARAM = SerialParam()? I agree that it can be frustrating to figure out the source of such errors and that the error seemingly being hidden doesn't help. — Reply to this email directly, view it on GitHub <#151 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABF2DHZCAOGIOYZGHVUPKN33LE43DAVCNFSM6AAAAACCT6KWJOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTCMZXHEZDAMRVHA> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

-- Best, Kasper