[BioC] WGCNA

[Martin Morgan]

On 02/16/2014 06:14 PM, Peter Langfelder wrote:
> As far as I know, RStudio has problems with parallel calculations that
> fork R processes (such as those called by parallel or
> foreach/doParallel on machines where fork is available). The WGCNA
> function pickSoftThreshold uses this parallelization. The correlation
> functions use multithreading from within C code, which does not fork
> the entire R session. That's why corFast etc in WGCNA work fine but
> pickSoftThreshold does not.

but from my original mail pickSoftThreshold 'worked for me' under Rstudio on 
Windows (output below, again); maybe I didn't invoke it in a way that would 
trigger the error, so a reproducible illustration would help (me). Also on 
Windows (platform of the original post) the parallel package doesn't fork, e.g., 
from ?mcfork

      These are low-level functions, not available on Windows, and not
      exported from the namespace.

or ?mclapply

      It relies on forking and hence is not available on Windows unless
      'mc.cores = 1'.

Do you mean something less literally forking, like spawning a new process? Maybe 
from

   http://stackoverflow.com/questions/20704235

the solution is as simple as adding .packages="WGCNA" to your foreach calls?

 > x = pickSoftThreshold(data)
trace: .C("checkAvailableMemoryForR", size = as.double(size), PACKAGE = "WGCNA")
trace: .C("corFast", x = as.double(x), nrow = as.integer(nrow(x)), ncolx = 
as.integer(ncol(x)),
     y = as.double(y), ncoly = as.integer(ncol(y)), quick = as.double(quick),
     cosineX = as.integer(cosineX), cosineY = as.integer(cosineY),
     res = as.double(bi), nNA = as.integer(nNA), err = as.integer(err),
     nThreads = as.integer(nThreads), verbose = as.integer(verbose),
     indent = as.integer(indent), DUP = FALSE, NAOK = TRUE, PACKAGE = "WGCNA")
trace: .C("corFast", x = as.double(x), nrow = as.integer(nrow(x)), ncolx = 
as.integer(ncol(x)),
     y = as.double(y), ncoly = as.integer(ncol(y)), quick = as.double(quick),
     cosineX = as.integer(cosineX), cosineY = as.integer(cosineY),
     res = as.double(bi), nNA = as.integer(nNA), err = as.integer(err),
     nThreads = as.integer(nThreads), verbose = as.integer(verbose),
     indent = as.integer(indent), DUP = FALSE, NAOK = TRUE, PACKAGE = "WGCNA")
    Power SFT.R.sq  slope truncated.R.sq  mean.k. median.k.   max.k.
1      1    0.173 -11.00        0.98400 4.02e+01  4.02e+01 4.45e+01
2      2    0.326  -8.45        0.95700 5.04e+00  5.04e+00 6.12e+00
3      3    0.241  -4.30        0.95300 8.01e-01  8.02e-01 1.05e+00
4      4    0.348  -4.04        0.95000 1.49e-01  1.49e-01 2.25e-01
5      5    0.513  -3.74        0.90400 3.15e-02  3.09e-02 5.77e-02
6      6    0.697  -3.72        0.92600 7.32e-03  6.99e-03 1.70e-02
7      7    0.811  -3.26        0.93300 1.85e-03  1.70e-03 5.55e-03
8      8    0.893  -2.92        0.95700 4.99e-04  4.32e-04 1.97e-03
9      9    0.923  -2.67        0.94400 1.43e-04  1.15e-04 7.35e-04
10    10    0.958  -2.38        0.96500 4.34e-05  3.15e-05 2.85e-04
11    12    0.910  -2.07        0.88600 4.50e-06  2.59e-06 4.55e-05
12    14    0.325  -3.02        0.21800 5.31e-07  2.29e-07 7.58e-06
13    16    0.219  -3.01        0.00942 6.88e-08  2.06e-08 1.29e-06
14    18    0.206  -3.06        0.03550 9.53e-09  1.93e-09 2.29e-07
15    20    0.227  -2.97        0.02620 1.39e-09  1.85e-10 4.11e-08




>
> Hope this clears up any lingering confusion.
>
> Also see inline below
>
> On Sun, Feb 16, 2014 at 12:26 PM, Martin Morgan <mtmorgan at fhcrc.org> wrote:
>
>>
>>
>> Also I don't see where the error message is coming from (who is printing
>> "task 1 failed - " ?)
>
>
> I believe this is printed by foreach/doParallel but here I could be wrong.
>
>
>
> HTH,
>
> Peter
>
>>
>> I have a vague recollection that the x[["Version"]] (which comes from
>> getAnywhere("print.sessionInfo")) has to do with multiple installed versions
>> of a package, but again it would be good to get to the bottom of this
>> problem.
>>
>> Martin


-- 
Computational Biology / Fred Hutchinson Cancer Research Center
1100 Fairview Ave. N.
PO Box 19024 Seattle, WA 98109

Location: Arnold Building M1 B861
Phone: (206) 667-2793