[BioC] Limma for GoldenGate Methylation Cancer Panel I

[Chao-Jen Wong]

Hi, Jinyan

I believe limma is suitable for GoldenGate Cancer Panel data assuming
you are using an appropriate normalization method and the design
matrix.  Perhaps you should consider using the 'methylumi' package to
preprocess (QA and normalization) your data before proceeding the
differential methylation analysis.

Chao-Jen

On 07/06/10 06:39, Sean Davis wrote:
> On Tue, Jul 6, 2010 at 9:31 AM, James W. MacDonald <jmacdon at med.umich.edu>wrote:
>
>   
>> Hi Jinyan Huang,
>>
>> On 7/6/2010 4:08 AM, Jinyan Huang wrote:
>>
>>     
>>> Does anyone have the exprience to use limma for two-color array:
>>> GoldenGate Methylation Cancer Panel I (Golden Gate Cancer Panel
>>> Methylation Illumina)
>>>
>>> I used it to analysis Methylation data for finding the different
>>> methlated genes, but the result is not good. There are too many small
>>> p-value the result. there are biology repeat in my data. My R code is
>>> like this:
>>>
>>> library(limma)
>>> exp<-read.table("exp.txt",F)
>>> sample_id<-read.table("sample_id",F)
>>> row.names(exp)<-exp[,1]
>>> exp<-exp[,-1]
>>> design<-read.table("design.txt",F)
>>> explow<-exp[,design[1,]==-1]
>>> exphigh<-exp[,design[1,]==1]
>>> expsort<-cbind(explow,exphigh)
>>> idlow<-sample_id[design[1,]==-1]
>>> idhigh<-sample_id[design[1,]==1]
>>> idsort<-c(idlow,idhigh)
>>> colred<-rep("red",length(exphigh[1,]))
>>> collow<-rep("blue",length(explow[1,]))
>>> col<-c(collow,colred)
>>> MA<-as.matrix(expsort)
>>> exp_norm<-normalizeBetweenArrays(MA,method="scale")
>>>
>>>       
>>     
> And this normalization is almost certainly not going to get you what you
> want, assuming they are the "beta" values from Illumina.
>
> Sean
>
>
>
>   
>> design_sort<-c(rep(-1,length(collow)),rep(1,length(colred)))
>>     
>>>       
>> Wow. That's a lot of code to end up with a two-column matrix consisting of
>> a column of -1s and a column of 1s. Is there some reason that modelMatrix()
>> doesn't do what you want?
>>
>> I also suspect that the design matrix you came up with isn't correct for
>> your experiment. I can't envision how the design matrix you have makes any
>> sense. But without knowing the experimental design, I can't say for sure.
>>
>> I would recommend finding a local statistician who might be able to help
>> you with this analysis.
>>
>> Best,
>>
>> Jim
>>
>>
>>  fit<- lmFit(MA,design_sort)
>>     
>>> fit<- eBayes(fit)
>>> mylist<-topTable(fit,number=Inf,adjust="BH")
>>>
>>> _______________________________________________
>>> Bioconductor mailing list
>>> Bioconductor at stat.math.ethz.ch
>>> 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
>> Douglas Lab
>> University of Michigan
>> Department of Human Genetics
>> 5912 Buhl
>> 1241 E. Catherine St.
>> Ann Arbor MI 48109-5618
>> 734-615-7826
>> **********************************************************
>> Electronic Mail is not secure, may not be read every day, and should not be
>> used for urgent or sensitive issues
>> _______________________________________________
>> Bioconductor mailing list
>> Bioconductor at stat.math.ethz.ch
>> https://stat.ethz.ch/mailman/listinfo/bioconductor
>> Search the archives:
>> http://news.gmane.org/gmane.science.biology.informatics.conductor
>>
>>     
> 	[[alternative HTML version deleted]]
>
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/bioconductor
> Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor
>   


-- 
Chao-Jen Wong
Program in Computational Biology
Division of Public Health Sciences
Fred Hutchinson Cancer Research Center
1100 Fairview Avenue N., M1-B514
PO Box 19024
Seattle, WA 98109
206.667.4485
cwon2 at fhcrc.org