[BioC] Utilizing outlier information in RMA and MAS estimation?

[Lishuang Shen]

Dear Laurent and Ben:

Thank you for the quick and instructive replies. 

Yes, most CEL files has very little fraction of  outlier cell, so it should not 
cause much difference. Just that I came across a data set, with >5000 cells 
labled as outlier (>2% of total cells). Then I thought it might make difference 
with outlier information for it.

I will evaluate the need and decide if take outliers into consideration in 
analysis or to do it someway.

Regards,

Lishuang

> The handling of NAs in not implemented for all the processing methods,
> unfortunately. We had some discussions about that around the
> first release, agreed that it would be cool to have it for all,
> but ... it never went top priority since few appear to use the
> feature (or may be few do so 'cause it's not working in every case).
> Other may correct me, I think that skipping the background correction
> for now should make it through (for mas, may be for rma too).
> If you really-really-really what it to work and come with a fix,
> it would be neat too...
> 
> Sorry, not much more to suggest for now...
> 
> 
> 
> Hopin' it helps,
> 
> 
> 
> Laurent
> 
> 
> 
> 
> 
> On Wed, Jul 16, 2003 at 02:35:05PM -0500, Lishuang Shen wrote:
> > Hi:
> > 
> > With ReadAffy, Affy can mask the outliers as NA with option: 
rm.outlier=TRUE.
> > 
> > But I could not do RMA or MAS 5 estimation with AffyBatch obtained in this
way. 
> > either using rma or using expresso, as I always got the error message about 
> > NA/NaN:
> > 
> > >background correction: mas 
> > >PM/MM correction : mas 
> > >expression values: mas 
> > >background correcting...Error in as.vector(data) : NA/NaN/Inf in foreign 
> > function call (arg 1)
> > 
> > Has anyone used outlier information during the expression estimation? I 
> > searched all the mailing list archieve, but found no mention on if this 
> > information is use or not, or information on if outliers marked as NA 
allowed 

> > in estimation.
> > 
> > 
> > Thank you for your time.
> > 
> > 
> > Lishuang Shen
> > 
> > BarleyBase,
> > Iowa State University
> > Ames, Iowa 50010
> > Tel: (515)-294-8479 (Office
> > Email: lshen at iastate.edu; shen_lishuang at yahoo.com
> > 
> > _______________________________________________
> > Bioconductor mailing list
> > Bioconductor at stat.math.ethz.ch
> > https://www.stat.math.ethz.ch/mailman/listinfo/bioconductor
> 
>