[BioC] Agilent miRNA one color platform analysis

Lauro Sumoy Van Dyck lauro.sumoy at crg.es
Thu Feb 7 19:03:59 CET 2008

Dear Francesco,

I will follow up on your e-mail regarding Agilent miRNA single channel
microarray normalization. As I had mentined at the moment we are working
with genepix data, although we may play around with Agilent feature
extraction as well we feel more comfortable with managing genepix
extracted gpr files with the irregular feature finding option.

I am able to load data from single channel gpr files by asking it to
load the green channel data twice.

RG <- read.maimages(targets,source="genepix", columns=list(R="F532
Median",G="F532 Median",Rb="B532 Mean",Gb="B532 Mean"))
vG <- normalizeBetweenArrays(RG$G,method="vsn")

Regarding use of normexp along with vsn, the limma manual says vsn
should be applied to non-background subtracted data. Even so I tried do
do this.

After applying normexp to single channel data (no offset applied)
followed by vsn, the resulting log2intensities are way off (no values
under 12!!!):

BSubRG <- backgroundCorrect(RG, method="normexp")
vG <- normalizeBetweenArrays(BSubRG$G,method="vsn")

What offset values do you suggest using when applying normexp background

Do you actually use normexp along with vsn or is it that you apply
normexp along with quantile normalization or vsn on its own?

Thanks again for your input. Other people's input would be appreciated.


Lauro Sumoy
Microarray Laboratory
Bioinformatics and Genomics Program
Center for Genomic Regulation (CRG)
Dr. Aiguader 88
08003 Barcelona
Office Phone: +34-93-316-0125

CRG Fax: +34-93-316-0099

e-mail: lauro.sumoy at crg.es

-----Mensaje original-----
De: bioconductor-bounces at stat.math.ethz.ch
[mailto:bioconductor-bounces at stat.math.ethz.ch] En nombre de Francesco
Enviado el: lunes, 04 de febrero de 2008 12:32
Para: Bioconductor
Asunto: Re: [BioC] Agilent miRNA one color platform analysis

Dear Lauro,

I'm happy some other people is interested in Agilent miRNA microarrays.

We are using this platform for some time now, and I believe I managed to
have some good results with the limma package.

If you are using the Feature Extraction from Agilent you should just
care a good function to weighting the spots. If you read the FE manual
you see you have two parameters WellAboveBG and isPosAndSignif.
Personally I prefer isPosAndSignif, because is less restrictive (Well
above is isPosAndSignif plus other tests...).

Than you have to pay attention to the "one-channel" problem with the
Agilent microarray in limma. you can find some post in the mailing list
about this matter.

Background Correction we are using normexp, with an adequate offset. Is
proved that normexp is a good method and it's suitable in our case, in
fact in miRNA we have a lot of low intensity spots, normexp + offset fix
it if this is the case.

Than the normalisation.. this is the most problematic part. We find out
that VSN is the best normalisation for us.

I suggest you to have a look at :

Davidson T.S., Johnson C.D. and Andruss B.F. "Analyzing micro-RNA
expression using microarrays" Methods in Enzymology 411(1):14-34, 2006.

Best regards


	[[alternative HTML version deleted]]

Bioconductor mailing list
Bioconductor at stat.math.ethz.ch
Search the archives:

More information about the Bioconductor mailing list