[BioC] Bioconductor Digest, Vol 87, Issue 10

[Richard Pearson]

Hi Avinash

So, it looks from your code like you haven't provided any phenotype information about your 
CEL files, and therefore each CEL file will be treated as a different condition. Because 
you have only 1 array per condition (i.e. you have no replicates), limma is going to give 
an error. Please note the following line from the puma User Guide (bottom of page 4 if 
you're using puma 2.0.0, but you still haven't given me the output from sessionInfo() so I 
don't know what version you're using):
"The easiest way to supply phenotype information is in a text file that is loaded using
the phenotype parameter of the ReadAffy function". Unfortunately, my documentation is 
incorrect here - ReadAffy has no phenotype parameter! Instead it should say phenoData 
parameter.

As I suggested in my first reply, showing the output of pData(eset_rma) would confirm that 
the above is true. From reading ?ReadAffy, my guess is that pData(eset_rma) will give you 
a data.frame with a column called sample containing the numbers 1 to 6.

To provide phenotype data for your eset_rma object you could do something like the following:

pData(eset_rma) <- data.frame(AvinashCondition=c("Condition1", "Condition1", "Condition1", 
"Condition2", "Condition2", "Condition2"))

"contrast" is a standard statistical term. I've tried to explain this in section 4.6 of 
the puma manual, but you could also look at the limma user guide, or google this (e.g. to 
get here: http://en.wikipedia.org/wiki/Contrast_%28statistics%29), or ask a local 
statistician.

Please do read the posting guide - it can be found here:
http://www.bioconductor.org/docs/postingGuide.html

Best wishes

Richard


avinash gupta wrote:
> sir,
>       puma clustering method . i make the object of (sample6, Celfile) 
>  by this command
>  
> dat<-ReadAffy()
> 
> library(pumadata)
> 
>  >eset_mmgmos<-mmgmos(dat)
>  >eset_rma<-rma(dat)
> 
> after that i do "dentifying differentially expressed genes" method in 
> this method i use this command
>  >eset_comb <- pumaComb(eset_mmgmos)
>  > pumaDERes <- pumaDE(eset_comb)
>  > limmaRes <- calculateLimma(eset_rma)      ## in this command it show 
> the  error like:
> Error in ebayes(fit = fit, proportion = proportion, stdev.coef.lim = 
> stdev.coef.lim) : 
>   No residual degrees of freedom in linear model fits
> i don't understand it 
> plz solve this and mail me .
> 
> i more things
> in the code 
> toppumaDEIntGene<-topGenes(pumaDERes, contrast = 7)
> 
> i don't under stand the value of contrast, how it's value define ,in my 
> obj have 6 cel file ,so plz help me to define its value.

-- 
Dr Richard D Pearson                       richard.pearson at well.ox.ac.uk
Wellcome Trust Centre for Human Genetics   http://www.well.ox.ac.uk/~rpearson
University of Oxford                       Tel: +44 (0)1865 617890
Roosevelt Drive                            Mob: +44 (0)7971 221181
Oxford OX3 7BN, UK                         Fax: +44 (0)1865 287664