[BioC] Where to get BAM files for easyRNASeq human use case ALSO ANNOTATION

Wolfgang Huber whuber at embl.de
Fri Aug 17 14:35:01 CEST 2012


Hi Martin

thanks! I wasn't aware this worked now (and it wasn't immediately 
obvious when I tried this morning). So we can move ahead. I'll discuss 
with Simon and Alejandro about how to proceed (without major disruption 
or -traction)

Steve: you asked why. For DESeq, we have had applications e.g. with 
count data from mass spec (where there is sometimes no associated 
genomic interval), or from HiC (where there are typically two genomic 
intervals for each count). I'd like to keep the flexibility for dealing 
with these sorts of situations. And people might always have count 
tables where the coordinates were dropped - while not ideal, there is no 
reason why DE(X)Seq should refuse to work with these.

	Best wishes
	Wolfgang




Martin Morgan scripsit 08/17/2012 02:16 PM:
> On 08/17/2012 04:36 AM, Steve Lianoglou wrote:
>> Hi,
>>
>> On Friday, August 17, 2012, Wolfgang Huber wrote:
>>
>>>
>>>
>>> On 8/17/12 9:48 AM, Nicolas Delhomme wrote:
>>
>>
>> [snip]
>>
>> We'd be happy to add methods or converters from SummarizedExperiment to
>>> DESeq's CountDataSet and DEXSeq's ExonCountSet classes, presumably into
>>> these packages.
>>>
>>> The problem is the reverse direction: SummarizedExperiment insists on
>>> having (non-NA) ranges information (start, end, width), while this is
>>> not a
>>> restriction that would make sense to impose on count tables for DESeq or
>>> DEXSeq.
>
> I recently implemented support for no coordinates,
>
>  > m = matrix(0, 10, 5, dimnames=list(LETTERS[1:10], letters[1:5]))
>  > sx = SummarizedExperiment(m) ## etc;
>  > class(rowData(sx))
> [1] "GRangesList"
> attr(,"package")
> [1] "GenomicRanges"
>
> the rowData is a GRangesList of length nrow(m), and with all ranges with
> length 0.
>
> Martin
>
>>
>>
>> Interesting.
>>
>> I'm trying to think of why this restriction doesn't make sense for DESeq
>> and co's count tables but I'm drawing a blank.
>>
>> The counts in each row of the count table are surely coming from some
>> genomic locus, no?
>>
>> Are you thinking about thing like gene fusion events or something?
>>
>> Thanks,
>> -steve
>>
>>
>>
>
>


-- 
Best wishes
	Wolfgang

Wolfgang Huber
EMBL
http://www.embl.de/research/units/genome_biology/huber



More information about the Bioconductor mailing list