RStudio/BiocParallel warning cause execution to crash

[yonbar]

RStudio has a known issue that produces warnings when running a parallel process, which can be read here or here or here.
When attempting fitExtractVarPartModel() / dream() with parallel process, this causes an exception:

library("variancePartition")
data(varPartData)
form <- ~ Age + (1 | Individual) + (1 | Tissue) + (1 | Batch)
varPart <- fitExtractVarPartModel(geneExpr, form, info, BPPARAM = SnowParam(4))

The exception (might be a different package, other than stats):

Error in serialize(data, node$con) : 
  (converted from warning) 'package:stats' may not be available when loading
Error in serialize(data, node$con) : error writing to connection

The exception is thrown as run_lmm() sets the warning level to 2:

# catch warnings like errors, then reset to original value
optValue <- getOption("warn")
options(warn = 2)
on.exit(options(warn = optValue))

Of course, the problem stems from RStudio/BiocParallel and not from variancePartition, but still - terminating the execution due to a warning might be a bit harsh, and in addition - users might not be aware that variancePartition code terminates upon a warning.

In the mean time, there is a simple walkaround, using BPPARAM = SnowParam(exportglobals = F)

[GabrielHoffman]

Can you show sessionInfo()

[yonbar]

R version 4.3.2 (2023-10-31 ucrt)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 10 x64 (build 19044)

Matrix products: default


locale:
[1] LC_COLLATE=English_United States.utf8  LC_CTYPE=English_United States.utf8    LC_MONETARY=English_United States.utf8 LC_NUMERIC=C                          
[5] LC_TIME=English_United States.utf8    

time zone: America/New_York
tzcode source: internal

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

other attached packages:
[1] variancePartition_1.32.2 BiocParallel_1.36.0      limma_3.58.1             ggplot2_3.4.4           

loaded via a namespace (and not attached):
 [1] gtable_0.3.4        caTools_1.18.2      Biobase_2.62.0      lattice_0.22-5      numDeriv_2016.8-1.1 vctrs_0.6.5         tools_4.3.2         Rdpack_2.6          bitops_1.0-7       
[10] generics_0.1.3      parallel_4.3.2      pbkrtest_0.5.2      tibble_3.2.1        fansi_1.0.5         pkgconfig_2.0.3     Matrix_1.6-4        KernSmooth_2.23-22  lifecycle_1.0.4    
[19] stringr_1.5.1       compiler_4.3.2      gplots_3.1.3        statmod_1.5.0       munsell_0.5.0       RhpcBLASctl_0.23-42 codetools_0.2-19    lmerTest_3.1-3      snow_0.4-4         
[28] pillar_1.9.0        nloptr_2.0.3        tidyr_1.3.0         MASS_7.3-60         aod_1.3.2           iterators_1.0.14    boot_1.3-28.1       nlme_3.1-164        gtools_3.9.5       
[37] tidyselect_1.2.0    stringi_1.8.2       fANCOVA_0.6-1       mvtnorm_1.2-4       dplyr_1.1.4         reshape2_1.4.4      purrr_1.0.2         splines_4.3.2       grid_4.3.2         
[46] colorspace_2.1-0    cli_3.6.1           magrittr_2.0.3      utf8_1.2.4          broom_1.0.5         corpcor_1.6.10      withr_2.5.2         scales_1.3.0        backports_1.4.1    
[55] remaCor_0.0.16      matrixStats_1.1.0   lme4_1.1-35.1       rbibutils_2.2.16    EnvStats_2.8.1      rlang_1.1.2         Rcpp_1.0.11         glue_1.6.2          BiocGenerics_0.48.1
[64] rstudioapi_0.15.0   minqa_1.2.6         plyr_1.8.9          R6_2.5.1      

Edit: The issue was closed by mistake

[GabrielHoffman]

I'm working on a fix

[GabrielHoffman]

Thanks for the bug report and the fix. I now set exportglobals <- FALSE internally to address this, so the standard code will work on windows now:

library("variancePartition")
data(varPartData)
form <- ~ Age + (1 | Individual) + (1 | Tissue) + (1 | Batch)
varPart <- fitExtractVarPartModel(geneExpr, form, info, BPPARAM = SnowParam(4))

See v1.33.4 from devtools::install_github("GabrielHoffman/variancePartition")

I will push this to the BioC release branch

Best,
Gabriel