[BioC] segfault ReadAffy cause 'memory not mapped'

Brian D. Peyser PhD brian.peyser at nih.gov
Fri Aug 23 05:15:32 CEST 2013

On Thu, 2013-08-22 at 22:15 -0400, Matthew McCall wrote:
> Brian,
> I'm curious what you ended up doing with fRMA. Did you just use the
> default implementation?
> If you want something closer to RMA, you can obtain RMA-like
> expression estimates by creating your own custom frma vectors using a
> (in your case potentially fairly large) subset of the data. You can
> even do this several times to get a sense of how much the estimates
> depend on the subset used to create the frma vectors. This is
> implemented in the frmaTools package. There is also a paper that
> describes this:
> McCall MN and Irizarry RA (2011). Thawing Frozen Robust Multi-array
> Analysis (fRMA), BMC Bioinformatics, 12:369.
> Best,
> Matt
Hi Matt,

Yes, I did run with the default fRMA vectors, and I have considered
creating custom vectors. I just haven't had the time yet to go back and
work that out. I'm sure I'll have to try a number of subsets, both
randomly selected and systematically compared. Rafa Irizarry is the one
who suggested I try fRMA in the first place (I went to him with the
RMA/GCRMA segfaults before coming here), and I have seen that citation.

Also, I don't have a reason to dislike the fRMA results, other than they
"felt" slightly different, especially in some specific comparisons I had
already made between certain subsets. The reason I was concerned about
the segfault was because I never seemed to approach my total physical
RAM, so that made me think there had to be some kind of limitation or
bug in the code. My goal here is more to help improve the package than
to get my data analyzed--there are several methods I can use to
accomplish the analysis, and I've already started with fRMA.


Brian D. Peyser PhD
Special Assistant to the Associate Director
Office of the Associate Director
Developmental Therapeutics Program
Division of Cancer Treatment and Diagnosis
National Cancer Institute
National Institutes of Health
301-524-5587 (mobile)

More information about the Bioconductor mailing list