[Bioc-devel] Sample names on NChannelSet

[Wolfgang Huber]

Hi Martin

my understanding of NChannelSet is that the element matrices of the 
assayData should have the same column names, which are identical with 
the "sampleNames".

Or do you see a convincing use case for different column names?

  Best wishes
	Wolfgang


Martin Morgan wrote:
> Hi Benilton --
> 
> Benilton Carvalho <bcarvalh at jhsph.edu> writes:
> 
>> (apologies in case this is a duplicated message - the first email was
>> returned with an error)
>>
>> NChannelSet accepts the elements of assayData to have different
>> colnames. For example:
>>
>> setClass("myClass", contains="NChannelSet")
>> x <- y <- matrix(0, 5, 5)
>> colnames(x) <- paste("x", 1:5, sep="")
>> colnames(y) <- paste("y", 1:5, sep="")
>> (zz1 <- new("myClass", x=x, y=y))
>> (zz2 <- new("myClass", x=x, y=x))
>>
>> Note that the show method suggests that the sample names for zz1 is
>> x1, ..., x5 (identical for zz2).
>>
>> Also, sampleNames() returns a list:
>>
>>  > sampleNames(zz1)
>> $x
>> [1] "x1" "x2" "x3" "x4" "x5"
>>
>> $y
>> [1] "y1" "y2" "y3" "y4" "y5"
>>
>>  > sampleNames(zz2)
>> $x
>> [1] "x1" "x2" "x3" "x4" "x5"
>>
>> $y
>> [1] "x1" "x2" "x3" "x4" "x5"
>>
>> The fact that NChannelSet accepts different sampleNames also affects
>> subsetting:
>>
>> cols <- c("x1", "x4")
>> (zz1[, cols])  ## Error: subscript out of bounds
>> (zz2[, cols])
>>
>> Is this an intended behavior? I'd expect sample names to be the same
> 
> NChannelSet assayData is meant to have elements that have the same
> dimensions, but that might have different names.
> 
> In terms of subsetting, you're right that subsetting by names (or in
> general subsetting so that you select different columns from different
> assayData elements) is not implemented when the assayData elements
> have different sampleNames (note that zz1[,c(1, 4)] works). I could
> implement a method taking a list to use for subsetting. but I suspect
> that is not really what you're after?
> 
> Martin
> 
>> across elements of AssayData and sampleNames() to return a vector
>> (instead of a list - side-effect of eapply).
>>
>> Thank you,
>>
>> benilton
>>
>> _______________________________________________
>> Bioc-devel at stat.math.ethz.ch mailing list
>> https://stat.ethz.ch/mailman/listinfo/bioc-devel
> 


-- 

Best wishes
      Wolfgang

----------------------------------------------------
Wolfgang Huber, EMBL-EBI, http://www.ebi.ac.uk/huber