[BioC] Quantile Normalization on mice data

[Arne.Muller at aventis.com]

Hi Yen Lin,

I've read your posting from 11 Feb to the BioC list about the design of your
experiment with two "batches" of mice.

I've a similar experimental layout: mouse cell cultures treated with a drig
at different doses. The experiment was carried out three times under the same
experimental conditions but in  different laboratories.

>From a hierachcical clustering and a principle component analysis of the
cross-chip normalized probe sets I can clearly see that the batches cluster
together.

I've used affyPLM to shed some light into the experiment, however, I'm a bit
lost with the interpretation. I was wondering how in the end you analysed
your data.

Did you try to use your own full factorial model for fitPLM, e.g. in your
case
	
	PM ~ -1 + probe + stage + batch

I've tried this with my experiment but I'm not sure how to interpret the
output ... .

	kind regards,

	Arne

> -----Original Message-----
> From: bioconductor-bounces at stat.math.ethz.ch 
> [mailto:bioconductor-bounces at stat.math.ethz.ch]On Behalf Of 
> Yen Lin Chia
> Sent: 11 February 2004 07:09
> To: bioconductor at stat.math.ethz.ch
> Subject: [BioC] Quantile Normalization on mice data
> 
> 
> Hi,
> 
> I'm working on some mice data from two batches (experiment carried out
> different time), but what I'm interested is to compare the gene
> expression between two stages
> 
> Stage I:  2 mice from batch D and 1 mouse from batch E
> Stage II: 1 from batch D and 4 from batch E.
> 
> Two tissue samples are taken for each mouse, center of the 
> tumor and the
> rim of the tumor.  Thus, I have two set of results layout
> (above-mentioned).  My first thought is to normalize the rim 
> and core of
> the tumor separately, but ignoring the batch variation.  
> Wonder if this
> is a good apprach.
> 
> Will the batch variation be problematic, from box plots, you can group
> the plots into batches (regardless of stages).
> 
> I'm trying to estimate the batch effect by using affyPLM, 
> (i.e. dropping
> samples as a factor), is p-values computed along with the function?  I
> only see the estimates and standard error.  I'm new to R and
> bioconductor packages.
> 
> Thanks.
> 
> Yen Lin
> 
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at stat.math.ethz.ch
> https://www.stat.math.ethz.ch/mailman/listinfo/bioconductor
>