I just discovered a bug in BiocParallel +, version 1.4.0 when using nested lists and MulticoreParam (with BatchJobs it works fine): bplapply applies a function to each subelement of x elements:

library(BiocParallel)
param <- MulticoreParam(1)
register(param)

x <- list(a=list(1,2),b=list(3,5))

bplapply(x,I)
## $a
## [1] 1

## $b
## [1] 2

## $<NA>
## [1] 3

## $<NA>
## [1] 5

lapply(x,I)
## $a
## [[1]]
## [1] 1

## [[2]]
## [1] 2

## $b
## [[1]]
## [1] 3

## [[2]]
## [1] 5

My session info: 

> devtools::session_info()
Session info -------------------------------------------------------------------
setting  value                     
version  R version 3.2.2 (2015-08-14)
system   x86_64, linux-gnu         
ui       X11                       
language (EN)                      
collate  en_US.UTF-8               
tz       Europe/Berlin             
date     2015-12-01                  

Packages -----------------------------------------------------------------------
package        * version date       source      
BiocParallel   * 1.4.0   2015-11-18 Bioconductor
devtools         1.9.1   2015-09-11 CRAN (R 3.2.2)
digest           0.6.8   2014-12-31 CRAN (R 3.2.2)
futile.logger    1.4.1   2015-04-20 CRAN (R 3.2.2)
futile.options   1.0.0   2010-04-06 CRAN (R 3.2.2)
lambda.r         1.1.7   2015-03-20 CRAN (R 3.2.2)
memoise          0.2.1   2014-04-22 CRAN (R 3.2.2)

Is this a known error? Is it fixed in BiocParallel version 1.5.0?

Thanks!

Thanks for reporting the bug and providing an example. This was a special case where number of workers = 1; so not really parallel at all. In this case, the method should dispatch to SerialParam() which is now does. You may have noticed the same code with workers > 1 gave the correct results. When no workers are specified they are auto-detected with multicoreWorkers():

> bplapply(x, I, BPPARAM=MulticoreParam())
$a
[[1]]
[1] 1

[[2]]
[1] 2

$b
[[1]]
[1] 3

[[2]]
[1] 5

This is fixed in BiocParallel 1.4.1 in release and 1.5.2 in devel. Both versions are available immediately in svn or with biocLite() tomorrow noon-ish PST.

Valerie