[BioC] Limma/Affy - Include expression Intensity in Final 'decideTests' results - limm2annaffy

[James W. MacDonald]

Hi Atul,

The writeFit function simply outputs ALL of the probesets, along with 
statistics and annotation. If you just want the probesets that are 
significant in one or more contrast, you could subset using the 
results.default object.

Or add things manually. Your choice.

Best,

Jim



On 4/30/2013 3:05 PM, Atul Kakrana wrote:
> Hi Jim,
>
> You suggestion did exactly what I wanted but number of genes in output
> is different for both:
>
> results.default<- decideTests(fit2)
> write.fit(fit2,results.default,file='RESULTS2.tsv', sep = '\t')
>
> AND
>
> write.table(writeFit(fit2,"mouse4302", esetRMA.Reduced), "TestData.tsv",
> sep = '\t', quote = FALSE, row.names = FALSE)
>
> The decideTest gives total of ~16000 genes where as 'writeFit' gives
> ~20,000 genes.  I believe this is because of the default p-val cutoff in
> decideTests :
> decideTests(object,method="separate",adjust.method="BH",p.value=0.05,lfc=0)
>
> As 'writeFit' currently does not support such functions (like
> decideTests) I think it would be better to add expression intensities
> manually to the results.
>
> Thanks
>
> Atul
>
>
> On 04/30/2013 12:59 PM, James W. MacDonald wrote:
>> Hi Atul,
>>
>> You might then try writeFit(), which is probably more to your liking.
>> Something like
>>
>> write.table(writeFit(fit2, "mouse4302", esetRMA.Reduced),
>> "All_data.txt", sep = "\t", quote=FALSE, row.names = FALSE)
>>
>> Also please note that Excel has an unfortunate habit of changing any
>> gene symbol that looks like a date into a date, and this is
>> irreversible (so e.g., the 1Apr gene will be converted to 4/1/2013,
>> because that makes sense...).
>>
>> You want to import the data into Excel rather than opening, and in the
>> third step of the wizard ensure that you change the format of the
>> symbol column to text rather than general.
>>
>> Best,
>>
>> Jim
>>
>> On 4/30/2013 12:32 PM, Atul Kakrana wrote:
>>> Hi James,
>>>
>>> Thanks for the suggestion. By using limma2annaffy I am being able to
>>> get expression intensity in final results but the problem is that a
>>> separate file is generated for each comparison For Ex. if we have two
>>> time points and one control than it will generate two different files.
>>>
>>> I was thinking of some way to add expression intensity to the results
>>> from 'decideTests' that has all the comparisons at one place as that
>>> is important for time series comparison. I think one way to do so is
>>> just extract the expression intensities from normalized data and add
>>> to final result file manually using MS Excel.
>>>
>>> Besides that I see an error when 'text=TRUE' option is used in
>>> limma2annafyy:
>>>
>>>>   limma2annaffy(esetRMA.Reduced, fit2, design, adjust =
>>> "none",contrast.matrix, annotation(esetRMA), pfilt = 0.01, fldfilt=
>>> 1, text = TRUE, html =FALSE)
>>>
>>> You are going to output 3 tables,
>>> With this many genes in each:
>>>    1918 1950 2424
>>> Do you want to accept or change these values? [ a/c ]
>>> a
>>> Warning message:
>>> In chkPkgs(chip) :
>>>     The mouse4302.db package does not appear to contain annotation data.
>>>
>>> The error does not appears with just html output.
>>>
>>> Thanks
>>>
>>> Atul
>>>
>>>
>>> On 29-Apr-13 9:49 AM, James W. MacDonald wrote:
>>>> Hi Atul,
>>>>
>>>>
>>>> On 4/28/2013 11:55 PM, Atul Kakrana wrote:
>>>>> Hi All,
>>>>>
>>>>> I am using 'affy' and 'limma' to analyze affymetrix data.
>>>>>
>>>>> Simplified code:
>>>>>
>>>>> mydata<- ReadAffy()
>>>>> esetRMA<- rma(mydata)
>>>>> sampletype<- c('1','1','1','2','2','2','3','3','3','4','4','4')
>>>>> group<-factor(sampletype)
>>>>> design<- model.matrix(~0+group)
>>>>> fit<- lmFit(esetRMA, design)
>>>>> contrast.matrix<-makeContrasts(B-A,C-A,D-A, levels=design)
>>>>> fit2<- contrasts.fit(fit,contrast.matrix)
>>>>> fit2<- eBayes(fit2)
>>>>> results.default<- decideTests(fit2)
>>>>>
>>>>> I get the results without any problem. But what I need is
>>>>> expression intensity of probes across all samples to be included in
>>>>> the results. I tried:
>>>>> Intensity<- (2^(exprs(esetRMA)))
>>>>>
>>>>> But cannot fit data to the results
>>>>> fData(results.default)<- Intensity
>>>>>
>>>>> Could somebody please tell me how can include expression
>>>>> intensities in the 'decideTests' output?
>>>> The output from decideTests() is just a matrix of 0 and 1, so you
>>>> wouldn't want to append anything to that to begin with.
>>>>
>>>> You might take a look at limma2annaffy() in affycoretools, which
>>>> does essentially what I think you want.
>>>>
>>>> Best,
>>>>
>>>> Jim
>>>>
>>>>
>>>>> AK
>>>>>
>>>>> _______________________________________________
>>>>> Bioconductor mailing list
>>>>> Bioconductor at r-project.org
>>>>> https://stat.ethz.ch/mailman/listinfo/bioconductor
>>>>> Search the archives:
>>>>> http://news.gmane.org/gmane.science.biology.informatics.conductor

-- 
James W. MacDonald, M.S.
Biostatistician
University of Washington
Environmental and Occupational Health Sciences
4225 Roosevelt Way NE, # 100
Seattle WA 98105-6099