[BioC] Exclude individual probes before normalization

[Ben Bolstad]

It definitely does not increase memory usage. The subsetting of the data
is actually done at the R level (via a call using pm() accessor on an
AffyBatch with a subset argument). Given that you are pulling out a
smaller chunk of data from the AffyBatch and at the C level the code
tries to be fairly parsimonious about any additional memory it requires
I would also guess that it is less memory (this is speaking without
making any attempts to bench mark it).

Best,

Ben



On Mon, 2006-09-11 at 12:22 -0400, James W. MacDonald wrote:

> I have never needed to use the subset argument, so don't know for sure. 
> However, the subsetting appears to be done at the C level rather than at 
> the R level, so I would bet there isn't much copying being done, which 
> is the only way I know that more memory would be required. My bet is 
> that it would take less memory.
> 
> Ben Bolstad would know for sure, since he wrote the C code.

> Jenny Drnevich wrote:
> > Do you know if using the 'subset' argument of rma() has any effect on the 
> > memory it uses? I'm not sure if it would decrease the memory requirements 
> > by working with less data, or actually increase it by doing a subsetting 
> > step somewhere.
> > 

-- 
Ben Bolstad <bmb at bmbolstad.com>
http://bmbolstad.com