[BioC] Parallelizing Biostrings::pairwiseAlignment locally with multicore::mclapply problems?

Steve Lianoglou mailinglist.honeypot at gmail.com
Tue Apr 17 17:57:36 CEST 2012


Hi,

I see that pairwiseAlignment can be ||-ized using Rmpi, but I wrote a
simple wrapper to use multicore since I'm using one machine anyway[*].

Let's try this in the simplest possible way and assume I have
`pattern` and `subject` XStringSets that I want to align against
eachother and `mc.cores` is the number of cores I want to use:

split.idx <- sort(seq_along(pattern) %% mc.cores)
mc.output <- multicore:::mclapply(1:mc.cores - 1L, function(idx) {
  pairwiseAlignment(pattern[take], subject[take], ...)
})

Then collect results in a similar fashion to `mpi.collate.pairwiseAlignment`

Anyway, I'm getting these warnings in each ||-ized loop like so:

  select: Interrupted system call

All google hits of this warning seem to suggest a problem w/ each
process accessing a shared resource -- in particular a database
connection, but that can't be the case here.

Before I sink the next X arbitrary-time-units into this, I'm just
curious if anybody knows what might be raising this warning?

[*] to be honest, I'm a bit lazy to wade through the particulars of
setting Rmpi/slaves locally to just use the builtin mpi.* functions --
if someone has a reference to a handy tutorial that summarizes what's
required to use Rmpi to ||-ize stuff locally (on os x ;-), it'd be
greatly appreciated.

Abridged sessionInfo():

R version 2.15.0 (2012-03-30)
Platform: x86_64-apple-darwin9.8.0/x86_64 (64-bit)

other attached packages:
Rmpi_0.5-9
Biostrings_2.25.1


Thanks,
-steve

-- 
Steve Lianoglou
Graduate Student: Computational Systems Biology
 | Memorial Sloan-Kettering Cancer Center
 | Weill Medical College of Cornell University
Contact Info: http://cbio.mskcc.org/~lianos/contact



More information about the Bioconductor mailing list