Parallelize 'BiocParallel' functions

Henrik Bengtsson

+ =

The futurize package allows you to easily turn sequential code into parallel code by piping the sequential code to the futurize() function. Easy!

TL;DR

You can use futurize to make BiocParallel functions parallelize via any of the [parallel backends] supported by Futureverse, e.g.

library(futurize)
plan(multisession)
library(BiocParallel)

slow_fcn <- function(x) {
  message("x = ", x)
  Sys.sleep(0.1)  # emulate work
  x^2
}

xs <- 1:10
ys <- bplapply(xs, slow_fcn) |> futurize()

Introduction

This vignette demonstrates how to use this approach to parallelize functions such as bplapply(), bpmapply(), and bpvec() in the BiocParallel package. For example, consider the bplapply() function. It works like base-R lapply(), but uses the BiocParallel framework to process the tasks concurrently. It is commonly used something like:

library(BiocParallel)
xs <- 1:1000
ys <- bplapply(xs, slow_fcn)

The parallel backend is controlled by the BiocParallel::register(), similar to how we use future::plan() in Futureverse. We can use the futurize package to tell BiocParallel to hand over the orchestration of parallel tasks to Futureverse. All we need to do is to pass the expression to futurize() as in:

library(BiocParallel)

library(futurize)
plan(multisession) ## parallelize on local machine

xs <- 1:1000
ys <- bplapply(xs, slow_fcn) |> futurize()
#> x = 1
#> x = 2
#> x = 3
#> ...
#> x = 10

Note how messages produced on parallel workers are relayed as-is back to the main R session as they complete. Not only messages, but also warnings and other types of conditions are relayed back as-is. Likewise, standard output produced by cat(), print(), str(), and so on is relayed in the same way. This is a unique feature of Futureverse - other parallel frameworks in R, such as parallel, foreach with doParallel, and BiocParallel, silently drop standard output, messages, and warnings produced on workers. With futurize, your code behaves the same whether it runs sequentially or in parallel: nothing is lost in translation.

The built-in multisession backend parallelizes on your local computer and it works on all operating systems. There are other parallel backends to choose from, including alternatives to parallelize locally as well as distributed across remote machines, e.g.

plan(future.mirai::mirai_multisession)

and

plan(future.batchtools::batchtools_slurm)

Supported Functions

The futurize() function supports parallelization of all BiocParallel functions that take argument BPPARAM. Specifically,

• bplapply() and .bplapply_impl()
• bpmapply() and .bpmapply_impl()
• bpvec()
• bpaggregate()

The following functions are currently not supported:

• bpiterate() - technically supported, but because this function does not support using DoparParam() with it, it effectively does not work with futurize()
• bpvectorize()
• register()

Bioconductor packages using BiocParallel

Most Bioconductor packages that support parallelization do so via BiocParallel internally. These packages typically expose a BPPARAM argument in their functions, which controls the parallel backend used. For example, DESeq2::DESeq() has a BPPARAM argument that defaults to BiocParallel::bpparam(), which corresponds to the currently registered BiocParallel backend. This means that, in order to parallelize such a function, one can call BiocParallel::register() to set a parallel backend, and then the function will use it automatically.

However, not all packages default to bpparam(). For example, sva::ComBat() defaults to bpparam("SerialParam"), which means it always runs sequentially unless you explicitly pass a parallel BPPARAM argument. Because of this, one cannot count on bpparam() being the default everywhere - some functions require an explicit BPPARAM to parallelize. With futurize, this is handled automatically: futurize() injects the appropriate BPPARAM argument regardless of what the default is, so that the parallel execution is performed via the Futureverse, where the parallel backend is controlled by future::plan().

Progress Reporting via progressr

For progress reporting, please see the progressr package. It is specially designed to work with the Futureverse ecosystem and provide progress updates from parallelized computations in a near-live fashion. See the vignette("futurize-11-apply", package = "futurize") for more details and an example.