Re: Interspecies differential expression of orthologs with Edger

Probably wrong, but the reason I thought of using quantile normalization is
the need to correct both for the species-length, and library size.


On Wed, Aug 27, 2014 at 2:40 AM, Gordon K Smyth <smyth-QpO9qclcJEu6c6uEtOJ/[email protected]> wrote:

> That doesn't look helpful to me.  I suggested that you incorporate gene
> lengths into the offsets, not do quantile normalization of cpms.
>
> Sorry, I just don't have time to develop a code example for you.  I hope
> someone else will help.
>
> The whole topic of interspecies differential expression is a can of worms.
> Even if you adjust for gene length, there will still be differences in gc
> content and mappability between the species.
>
> Gordon
>
>
> On Wed, 27 Aug 2014, assaf www wrote:
>
>  Dear Gordon thanks,
>>
>> Suppose I start with the following matrices:
>>
>> # 'counts' is the Rsem filtered counts
>>
>>> counts[1:4,]
>>>
>>                     h0  h1  h2  n0  n1  n2
>> ENSRNOG00000000021   36  17  20  10  25  38
>> ENSRNOG00000000024 1283 615 731 644 807 991
>> ENSRNOG00000000028   26  12  11  18  23  28
>> ENSRNOG00000000029   22  13  12  16  17  15
>>
>> # 'geneLength' is the species-specific gene lengths, for species 'h' and
>> 'n':
>>
>>> geneLength[1:3,]
>>>
>>                   h0.length h1.length h2.length n0.length n1.length
>> n2.length
>> ENSRNOG00000000021      1200      1200      1200      1303      1303
>> 1303
>> ENSRNOG00000000024      1050      1050      1050      3080      3080
>> 3080
>> ENSRNOG00000000028      1047      1047      1047      1121      1121
>> 1121
>>
>>
>> does the following code look correct, and may allow allows across species
>> analysis ?:
>> (technically it works, I checked)
>>
>> # quantile normalization: (as in here:
>> http://davetang.org/wiki/tiki-index.php?page=Analysing+
>> digital+gene+expression
>> )
>>
>> x1 = counts*1000/geneLength
>> library(limma)
>> x2 = normalizeBetweenArrays(data.matrix(x1),method="quantile")
>> offset = log(counts+0.1)-log(x2+0.1)
>>
>> ...
>>
>> y <- estimateGLMCommonDisp(y,design,offset=offset)
>> y <- estimateGLMTrendedDisp(y,design,offset=offset)
>> y <- estimateGLMTagwiseDisp(y,design,offset=offset)
>> fit <- glmFit(y,design,offset=offset)
>>
>> ...
>>
>>
>> Thanks in advance for any help..,
>> Assaf
>>
>>
> ______________________________________________________________________
> The information in this email is confidential and inte...{{dropped:10}}

_______________________________________________
Bioconductor mailing list
[email protected]
https://stat.ethz.ch/mailman/listinfo/bioconductor
Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor