<html>
<body>
If I am doing comparisons across treatments, anything that is done
differently to normalized data receiving different treatments makes me
very nervous because it will contribute to the "treatment
difference". On the other hand, if you are using a randomized
<b>complete</b> block design, it would be OK to normalize within block,
because the normalization would affect every treatment equally.<br><br>
--Naomi<br><br>
At 04:50 AM 3/26/2004, Arne.Muller@aventis.com wrote:<br>
<blockquote type=cite class=cite cite>Hello,<br><br>
For a factorial design with some replicates at each level one could
assume<br>
that the replicates within each level are more similar to each other
than<br>
across levels (or even factors, especially if the factor is something
like<br>
"patient" or "animal"). <br><br>
One could normalize within the replicates 1st and then a 2nd round
of<br>
normalizaton across the factor levels. Would this make sense (at
all)?<br><br>
Has anybody tried this with bioC methods such as rma+quantile? I'd
be<br>
interested in your comments about this topic.<br><br>
<x-tab> </x-tab>kind
regards,<br><br>
<x-tab> </x-tab>Arne<br><br>
--<br>
Arne Muller, Ph.D.<br>
Toxicogenomics, Aventis Pharma<br>
arne dot muller domain=aventis com<br><br>
_______________________________________________<br>
Bioconductor mailing list<br>
Bioconductor@stat.math.ethz.ch<br>
<a href="https://www.stat.math.ethz.ch/mailman/listinfo/bioconductor" eudora="autourl">https://www.stat.math.ethz.ch/mailman/listinfo/bioconductor</a></blockquote>
<x-sigsep><p></x-sigsep>
Naomi S.
Altman
814-865-3791 (voice) <br>
Associate Professor <br>
Bioinformatics Consulting Center<br>
Dept. of
Statistics
814-863-7114 (fax) <br>
Penn State
University
814-865-1348 (Statistics) <br>
University Park, PA 16802-2111<br>
</body>
</html>