[BioC] Limma_Agilent_2color_Array

Wolfgang Huber huber at ebi.ac.uk
Wed Apr 16 19:37:48 CEST 2008


Dear Abilash,

you could try to run the arrayQualityMetrics report from the eponymous
package (and preferably a current devel version [1]). The plots in there
would help you to assess both whether there are gross problems in the
data, and how much and what type of normalisation is needed.

Adequate normalisation removes unwanted technical variation while it
maintains the interesting biological signal in the data. The best way to
assess whether this is goal is achieved is by looking at the behaviour
of the controls that were done as part of the experiment.

[1]
http://www.bioconductor.org/packages/2.2/bioc/html/arrayQualityMetrics.html

Best wishes
 Wolfgang

------------------------------------------------------------------
Wolfgang Huber  EBI/EMBL  Cambridge UK  http://www.ebi.ac.uk/huber


16/04/2008 18:02 Abhilash Venu a écrit
> Hi all,
> 
> I have a few questions. I am analyzing  20 experiments  performed  in
> Agilent 4x44K. Tumor and adjacent normal were labeled with Cy5 and Cy3
> respectively. The experiments include two dye swaps also. The feature
> extraction were performed using Agilent feature extraction software version
> 8.5. The txt files obtained after feature extraction, subjected to further
> analysis by limma. The script has given below
> txt_files <- dir(pattern=".txt")
> RG<-read.maimages(txt_files, source="agilent")
> design<-c(1,1,1,1,1,1,1,1,1,1,1,-1,1,1,1,1,1,1,1,-1) # this include two dye
> swap also
> Rgene<-backgroundCorrect(RG,method="normexp")
> fit<-lmFit(Rgene,design)
> fit<-eBayes(fit)
> topTable(fit,adjust="fdr", number=100)
> 
> As a result I get the format in which 'B' indicates what parameter? And is
> this analysis method is enough to get a good data?
> In Agilent arrays, I think generally it is normalized, in that case how the
> data will be affected by normalizing it again using above method?
> How can we decide which normalization methods should be used? Do you think
> its always better to decide by plotting different graphs?
> 
> Comments and answers will be appreciated.
> 
> Regards,
> Abhilash
> 
>



More information about the Bioconductor mailing list