[BioC] how do people handling outliers and masked oligos?

[Ben Bolstad]

What you are seeing is a reflection of the fact that some
preprocessing/summarization functions do not handle missing values well
or at all (masked cells are treated as missing values). 

Typically for RMA expression measures (which seems to what you are
aiming for) we do not use masks instead allowing the robustness of the
summarization procedure deal with problem probes.


Ben


On Wed, 2003-10-29 at 07:22, weiss wrote:
> Hi,
> 
> I think I have serious problems with   expresso()
> 
> I use option rm.mask=TRUE in ReadAffy()
> and then tried to use
> expresso (daten, bgcorrect.method="rma",normalize.method="quantiles",
> pmcorrect.method="pmonly",summary.method="medianpolish")
> and several other settings.
> 
> as stated yesterday I get thousands of Error messages
> and only NAs as an output (for ALL probesets)
> when I have masked just 2 oligos one from probe set 1000_at
> and one from probe set 1001_at.
> 
> e.g.
> MASKED part in my CEL-files look like
> [MASKS]
> NumberCells=2
> CellHeader=X    Y
> 399    560
> 399    561
> 
> 
> on the other hand it seems to make no difference on the  expression value
> of probe set 1000_at whether I mask zero, one or all 16 oligos,
> the expression value of 1000_at is always the same.
> 
> the MASKED part in my CEL-files looks  for "all 1000_at oligos masked" 
> like this:
> 
> [MASKS]
> NumberCells=16
> CellHeader=X    Y
> 399    560
> 544    186
> 530    506
> 617    350
> 459    490
> 408    546
> 484    312
> 548    334
> 578    370
> 498    466
> 503    442
> 482    440
> 397    546
> 352    466
> 253    496
> 228    632
> 
> has anyone countered the sample problem?
> since it seems to be same with outliers:
> how do bioconductor users handle OUTLIERS and MASKED oligos?
> 
> or am I completely wrong with what I intend to do?
> 
> gunter
> 
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at stat.math.ethz.ch
> https://www.stat.math.ethz.ch/mailman/listinfo/bioconductor
-- 
Ben Bolstad <bolstad at stat.berkeley.edu>