[BioC] Ringo package: error using computeRunningMedians function

[Joern Toedling]

Hi Diego,

I had a look at the data that you sent me. The problem indeed is that there
are some probes listed in the probeAnno object that do not correspond to any
features/probes of the ExpressionSet.

> table(ProbeAnno["1.index"] %in% featureNames(eSet)) # all TRUE
> table(ProbeAnno["10.index"] %in% featureNames(eSet)) # all TRUE
> table(ProbeAnno["11.index"] %in% featureNames(eSet)) # 20 FALSE

FALSE  TRUE
   20 16294

> ProbeAnno["11.index"][!(ProbeAnno["11.index"] %in% featureNames(eSet))]
 [1] "chr11.118393537.118393597" "chr11.118393665.118393725"
 [3] "chr11.118393916.118393976" "chr11.118393984.118394044"
  .... # lines removed
[19] "chr11.118394982.118395042" "chr11.118395017.118395077"

These are the chromosome-11 probes which do not appear in your ExpressionSet
'eSet'. Some of the other chromosomes also contain a couple of probes that are
not contained in eSet.

Therefore, your probeAnno object does not exactly correspond to your supplied
ExpressionSet.
In the development version of Ringo (v1.9.9), I have now added additional code
which reports a warning in such a case and ignores non-corresponding probes
from the probeAnno object.

You will be able to obtain this development version in a few days from:
http://www.bioconductor.org/packages/devel/bioc/html/Ringo.html

For now, you should consider rebuilding your probeAnno object such that it
matches the ExpressionSet eSet. For instance, you can remove all probes that
are not in the featureNames(eSet) from the POS-data.frame/file before building
the probeAnno object via 'posToProbeAnno'.

Regards,
Joern

On Mon, 31 Aug 2009 16:33:28 +0000, Diego Villar wrote
> Dear Joern,
> 
>   I looked at the output of the commands you suggested (in some instances
> only a fragment of the output is copied):
> 
> table(ProbeAnno["11.unique"])
> 
>     0
> 16314
> 
> table(ProbeAnno["11.index"])
> 
>      chr11.990402.990462       chr11.990469.990529      
> chr11.990589.990649
>                         1                         1                  
>       1      chr11.990598.990658   chr11.99098574.99098634 
chr11.99098696.99098756
>                         1                         1                  
>       1 
> Checking on the error as you indicated I retrieve the following values:
> 
> smoothX <- computeRunningMedians(eSet, probeAnno=ProbeAnno,
> + modColumn = "Cy5", winHalfSize = 400)
> 
> Chromosome 1 ...
> 170809ds ... 261108no ...
> Chromosome 10 ...
> 170809ds ... 261108no ...
> Chromosome 11 ...
> 170809ds ... Error en newExprs[chridx, i] <- chrrm :
>   NAs no son permitidos en asignaciones subscritas
> 
> Enter a frame number, or 0 to exit
> 
> 1: computeRunningMedians(eSet, probeAnno = ProbeAnno, modColumn =
> "Cy5", winHa
> 
> Selección: 1
> Called from: eval(expr, envir, enclos)
> Browse[1]> str(chridx)
>  int [1:16314] 40027 121370 99412 63143 131069 125479 54552 32661 81358
> 80190 ...
> Browse[1]> str(i)
>  int 1
> Browse[1]> str(chrrm)
>  Named num [1:16314] NA NA NA NA 0.164 ...
>  - attr(*, "names")= chr [1:16314] "186594" "186802" "186874"
> "186940" ...
> 
>   Thanks,
> 
>        Diego

---
Joern Toedling
Institut Curie -- U900
26 rue d'Ulm, 75005 Paris, FRANCE
Tel. +33 (0)156246926