[BioC] lmFit() with ExpressionSet

Gordon K Smyth smyth at wehi.EDU.AU
Mon Apr 8 01:05:26 CEST 2013


As James says, the statistical results are identical, but there's a 
possible difference in the annotation.  An ExpressionSet can contain probe 
annotation as well as expression values.  If it does, and you use

   fit <- lmFit(esetRMA, design)
   fit <- eBayes(fit)
   topTable(fit)

then limma will find the probe annotation and present it to you in the 
top-table.  On the other hand,

   fit <- lmFit(exprs(esetRMA), design)

throws away any probe annotation, except for the row names of the 
expression matrix.

In the code pipeline in Atul's original email, there was no probe 
annotation in the ExpressionSet, hence nothing was lost.

Another subtlety is that when you pass a matrix to lmFit(), it then has to 
try to guess whether the matrix contains log-ratios (from a two color 
array) or log-expression values (from a single channel array).  It is 
better to pass the full object so that no guessing is necessary.

Best wishes
Gordon

> Date: Sat, 06 Apr 2013 13:47:39 -0400
> From: "James W. MacDonald" <jmacdon at uw.edu>
> To: Atul Kakrana <atulkakrana at outlook.com>
> Cc: Atul Kakrana <atulkakrana at gmail.com>,	"bioconductor at r-project.org"
> 	<bioconductor at r-project.org>
> Subject: Re: [BioC] Problem with Limma - Possible Limma/affy bug -
> 	minimal code not working - Solved - Minor doubt
>
> Hi Atul,
>
> On 4/5/2013 7:08 PM, Atul Kakrana wrote:

>> What is the difference in using
>> fit<- lmFit(exprs(esetRMA), design)
>>
>> OR
>> fit<- lmFit(esetRMA, design)
>>
>> Does this effect result in any possible way?
>
> No. From ?lmFit:
>
> |object|
>
> object of class |numeric|, |matrix|, |MAList|, |EList|, |marrayNorm|, 
> |ExpressionSet| or |PLMset| containing log-ratios or log-values of 
> expression for a series of microarrays
>
> In the first case you are feeding lmFit a matrix, in the second an 
> ExpressionSet. Internally lmFit will then do exprs(esetRMA) to get the 
> data, so the results will be identical.
>
> Best,
>
> Jim

______________________________________________________________________
The information in this email is confidential and intend...{{dropped:4}}



More information about the Bioconductor mailing list