Goes somehow like that:

Create your covdesc tab-delimited file with the experiment description and CEL file names, as described 

http://www.bioconductor.org/packages/release/bioc/vignettes/simpleaffy/inst/doc/simpleAffy.pdf

Then in R:

library(simpleaffy)

raw.data

Are you looking for any specific dataset then let me know the dataset ID. There is a microarray collection I want to see if that dataset is available then I can go through with the steps.

First you have to format your data properly as suggested by Michal, then you could try BAGEL at http://www.yale.edu/townsend/software.html

Dear Marisol, Michal  and Zheng

Thank you very much. Your advice are very helpful for me!

Thanks All and Best Regard

Hi Jone, 

The easiest, though I might be biased, is to look for the GSExxx in InSilico DB, select the "Analysis" button and export to Gene-E. In Gene-E you can do a marker selection analysis to calculate p-values, fold-change, etc..

If the GSExxx is not in InSilico DB you have two options:

1) upload the .CEL files to InSilico DB

2)request the GSE (you will have to wait ~24h for it to be available).

If you prefer to use R for further analysis you can also download the ready-to-analyze data in R format or access it programmatically through the R-Bioconductor packages.

Here is a step-by-step tutorial to analyze .CEL or Excel files:

https://insilicodb.com/save-time-1-click-high-quality-annotations-and-genomic-data-in-gene-e/

Hope this helps,

Alain

Use limma package of bioconductor. You can read its manual.  It will give you exactly you wanted.  

The answers given so far will almost certainly solve your problem. But you still might want to have a look at our arrayanalysis.org. It does process the .cel files with the Bioconductor packages mentioned and a few more, it gives you quality control data and performs (different kinds of) normalisation and annotates the probesets. For the annotation you can choose between the native Affymetrix annotation or the alternative Brainarray custom cdf annotation that remapped probes to new genome builds to be sure that every probe really aligns with the target and also only hits that target. The new version can even do some basic statistics for you.