[BioC] edgeR GLM error

[Gordon K Smyth]

Dear Javier,

I don't see the slightest evidence of any error or problem.

You cannot prejudge how many DE transcripts you should get, and I have no 
idea what you mean by "associated transcripts" anyway.

This is the last reply I will make to this thread.

If you wish to receive help with using edgeR, please use a more 
informative subject heading for your email and explain your question in 
terms that others can understand.

Best wishes
Gordon

---------------------------------------------
Professor Gordon K Smyth,
Bioinformatics Division,
Walter and Eliza Hall Institute of Medical Research,
1G Royal Parade, Parkville, Vic 3052, Australia.
http://www.statsci.org/smyth

On Thu, 20 Sep 2012, Javier Simón-Sánchez wrote:

> HI, thanks a lot for your reply. The reason I think there's an error is
> because i get ~3000 associated transcripts, which in my opinion is too
> much, espescially because is the same amount I would get when using a
> pairwise comparison. Because of this, I think the package is not modulating
> for the brain region...
>
> I checked in the list but no one had this problem. Any ideas?
>
> Thanks a lot for our time and replies!
>
> On Thu, Sep 20, 2012 at 1:38 AM, Gordon K Smyth <smyth at wehi.edu.au> wrote:
>
>> Dear Javier,
>>
>> I don't see the slightest evidence of any error or problem.  In fact, you
>> say that everything is running properly.
>>
>> You might find it helpful to have a look at the posting guide:
>>
>>  http:://www.bioconductor.org/**doc/postingGuide.html<http://www.bioconductor.org/doc/postingGuide.html>
>>
>> Best wishes
>> Gordon
>>
>>
>>
>> On Wed, 19 Sep 2012, Javier Simón-Sánchez wrote:
>>
>>  Hello Gordon,
>>>
>>> Again, thank you for your recent help on this. As I tyold you, I managed
>>> to
>>> un my analysis. However, I get the impression that the model is adjusting
>>> propoerly.
>>>
>>> I want to run Cases vs Controls controling for differences beteween brain
>>> fregions but IM getting ~300- associated p values after correcting for
>>> multiple testing. The code im using is:
>>>
>>> library(limma)
>>> library(edgeR)
>>> x <- read.delim("/myexpressiondata.**txt",row.names="Symbol")
>>> group <- factor
>>> (c(1,1,1,1,1,1,1,1,1,1,1,1,1,**1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,**
>>> 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,**0,0,0,0,0,0))
>>> d <- DGEList(counts=x,group=group)
>>> cpm.d <- cpm(d)
>>> d <- d[rowSums(cpm.d > 1) >= 4, ]
>>> d <- calcNormFactors(d)
>>> region <-
>>> factor(c("Caudate","Frontal","**Putamen","Temporal","Caudate",**
>>> "Frontal","Hippocampus","**Putamen","Temporal","Caudate",**
>>> "Frontal","Hippocampus","**Putamen","Temporal","Caudate",**
>>> "Frontal","Hippocampus","**Putamen","Temporal","Caudate",**
>>> "Frontal","Hippocampus","**Putamen","Temporal","Caudate",**
>>> "Frontal","Hippocampus","**Putamen","Temporal","Caudate",**
>>> "Frontal","Hippocampus","**Putamen","Temporal","Caudate",**
>>> "Frontal","Hippocampus","**Putamen","Temporal","Caudate",**
>>> "Frontal","Hippocampus","**Putamen","Temporal","Caudate",**
>>> "Frontal","Hippocampus","**Putamen","Temporal"))
>>> design <- model.matrix(~region+d$**samples$group)
>>> d <- estimateGLMCommonDisp(d, design,verbose=TRUE)
>>> d <- estimateGLMTrendedDisp(d, design)
>>> d <- estimateGLMTagwiseDisp(d, design)
>>> glmfit.tagwise.d <- glmFit(d,design,dispersion=d$**tagwise.dispersion)
>>> lrt.tagwise.d <- glmLRT(glmfit.tagwise.d)
>>> topTags(lrt.tagwise.d)
>>>
>>>
>>>
>>> Thanks!
>>> On Thu, Sep 13, 2012 at 2:43 AM, Gordon K Smyth <smyth at wehi.edu.au>
>>> wrote:
>>>
>>>  Dear Javier,
>>>>
>>>> This error has been discused a number of times on this list.  The
>>>> solution
>>>> is to upgrade edgeR to the current devel version.
>>>>
>>>> Also please see the Bioconductor posting guide:
>>>>
>>>>  http://www.bioconductor.org/****help/mailing-list/posting-****guide/<http://www.bioconductor.org/**help/mailing-list/posting-**guide/>
>>>> <http://www.**bioconductor.org/help/mailing-**list/posting-guide/<http://www.bioconductor.org/help/mailing-list/posting-guide/>
>>>>>
>>>>
>>>>
>>>> Best wishes
>>>> Gordon
>>>>
>>>>  Date: Tue, 11 Sep 2012 13:04:55 +0200
>>>>
>>>>> From: Javier Sim?n-S?nchez <simonsanchezj at gmail.com>
>>>>> To: bioconductor at r-project.org
>>>>> Subject: [BioC] edgeR GLM error
>>>>>
>>>>> Hello,
>>>>> My name is Javier Simón Seanchez and I'm a post-doc at the VUmc in
>>>>> Amsterdam.
>>>>>
>>>>> The reason of this e-mail is that im running edgeR in an expression
>>>>> dataset
>>>>> and getting the following error when calculating the GLM common
>>>>> dispersion:
>>>>>
>>>>> *Error in beta[k, ] <- betaj[decr, ] :
>>>>>  NAs are not allowed in subscripted assignments
>>>>> *
>>>>>
>>>>> Im running cases versus controls and I want to modulate for different
>>>>> tissues. How can I overcome this error?
>>>>>
>>>>> Thanks a lot in advance
>>>>>
>>>>> --
>>>>> Javier Simon-Sanchez
>>>>>
>>>>
>>> --
>>> Javier Simon-Sanchez
>>>
>>
>> ______________________________**______________________________**__________
>> The information in this email is confidential and intended solely for the
>> addressee.
>> You must not disclose, forward, print or use it without the permission of
>> the sender.
>> ______________________________**______________________________**__________
>>
>
>
>
> -- 
> Javier Simon-Sanchez
>

______________________________________________________________________
The information in this email is confidential and intend...{{dropped:5}}