[Bioc-devel] [biocpkgs] Bug? in AssayData

[Martin Morgan]

Most of my last email was incorrect to one degree or another, so
thanks for your patience. I changed assayDataElementNames to use
storageMode(obj) == "list", moved the assayDataElement* functions to
methods-eSet, and changed the documentation to reflect the correct way
these functions should be used. All available in Biobase 1.11.42.

There are benefits to making these functions generics, and I'll make
changes in that direction shortly. Different options we are assessing
include (1) making 'names' generic,(2) making existing functions
assayDataElementNames etc generic, or (3) introducing new generics
e.g., assayNames and deprecating existing functions (not for the 2.4.0
release, though). Each of these has pros and cons, so input welcome.

Martin

Seth Falcon <sfalcon at fhcrc.org> writes:

> "Oosting, J. (PATH)" <J.Oosting at lumc.nl> writes:
>> My issue is with assayDataElementNames(), I just stumbled across
>> assayDataElement when I checked an error that popped up recently.
>>
>> I have an eSet subclass that has optional elements in assayData
>>
>> I use something like if ( !("intensity" %in%
>> assayDataElementNames(object))) object<-CalculateIntensity(object)
>>
>>
>> When the object has storageMode=="list" I get an error, because the
>> function assayDataStorageMode tests this assuming the argument is an
>> AssyData instance, but the following commands in the function assume
>> this is an eSet instance.
>>
>>> assayDataElementNames <- function(obj)   if
>>> (assayDataStorageMode(obj) == "list") names(assayData(obj)) else
>>> ls(assayData(obj))
>>
>> This should either be: if (assayDataStorageMode(assayData(obj)) ==
>> "list") or if (storageMode(obj) == "list")
>
> I agree there is ambiguity here that needs fixing.  Thanks for
> clarifying the issues you encountered.
>
> We will have some discussion among the developers here and have some
> sort of a fix posted asap.
>
> My initial feeling is that much of this code would be less confusing
> if these were methods and not functions.  
>
> Best,
>
> + seth
>
> _______________________________________________
> Bioc-devel at stat.math.ethz.ch mailing list
> https://stat.ethz.ch/mailman/listinfo/bioc-devel

-- 
Martin T. Morgan
Bioconductor / Computational Biology
http://bioconductor.org