[BioC] crlmm and snpMatrix

[Tim Rayner]

Hi,

I'm not sure if this really qualifies as a bug, but I've spent the
morning tracking down an odd interaction between the crlmm and
snpMatrix packages and I thought I should share it and maybe save
others some time. It looks as though an object in the snpMatrix
namespace is clobbering something needed by the crlmm2() call. When I
run the following, crlmm2 runs indefinitely, slowly consuming memory
until I eventually kill it:

# celdir is a directory containing cel files.
nclust=4
clusttype='MPI'
require(ff)
require(snow)
require(crlmm)
require(snpMatrix)   # comment this line out to fix
options(cluster=makeCluster(nclust, clusttype))
cels <- list.celfiles(celdir, full.names=TRUE)
result <- crlmm2(cels, verbose=TRUE)

If I comment out the "require(snpMatrix)" line then everything runs fine.

The typical output from hanging run looks like this:

Loading required package: genomewidesnp6Crlmm
Welcome togenomewidesnp6Crlmmversion 1.0.2
Loading annotations and mixture model parameters.
Initializing objects.
Processing 64 files.
Loading annotations.
Determining gender.
Calling 906600 SNPs for recalibration...

And finally, the sessionInfo():

R version 2.11.0 Patched (2010-04-30 r51866)
x86_64-apple-darwin9.8.0

locale:
[1] en_GB.UTF-8/en_GB.UTF-8/C/C/en_GB.UTF-8/en_GB.UTF-8

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

other attached packages:
[1] snpMatrix_1.12.0    survival_2.35-8     crlmm_1.6.0
[4] oligoClasses_1.10.0 Biobase_2.8.0       snow_0.3-3
[7] ff_2.1-2            bit_1.1-4

loaded via a namespace (and not attached):
 [1] affyio_1.16.0         annotate_1.26.0       AnnotationDbi_1.10.0
 [4] Biostrings_2.16.0     DBI_0.2-5             ellipse_0.3-5
 [7] genefilter_1.30.0     IRanges_1.6.6         mvtnorm_0.9-9
[10] preprocessCore_1.10.0 RSQLite_0.8-4         xtable_1.5-6

So to conclude: don't load snpMatrix until after you're done with
crlmm2. I'm afraid I've no idea what the actual cause of this problem
is, since I don't get any "The following object(s) are masked from"
messages when loading snpMatrix.

Best regards,

Tim Rayner

-- 
Bioinformatician, Smith Lab
CIMR, University of Cambridge