Hi All,
somehow from the back of my mind I think that this was already
discussed here
aoms time ago, but I cannot find the postings, os please excuse if
this is a
duplication ... .
When processing some chips with RMA background correction, crooss chip
quantile normalisation and "pmonly" probe set summary, the
distribution of
the intensities per chip is not normal.
I haven't tried
--
Arne Muller, Ph.D.
Toxicogenomics, Aventis Pharma
arne dot muller domain=aventis com
Sorry, my last posting was incomplete (slipped over the keyboard ...).
I meant that I haven't explored other methods yet, but since the RMA
values
are log2, I thought that I'd get something close to a normal
distribution.
Comapred to a normal distribution I get many low intensity probe sets.
The values are generated like this:
eset.rma <- expresso(cel, bgcorrect.method="rma",
normalize.method="quantiles", pmcorrect.method="pmonly",
summary.method="medianpolish")
then:
hist(exprs(eset.rma[,10]))
kind regards,
Arne
--
Arne Muller, Ph.D.
Toxicogenomics, Aventis Pharma
arne dot muller domain=aventis com
Arne,
I'd be very surprised if you'd observe a normal distribution. You
should not
rely on this assumption! I haven't seen any data yet that looked like
normal
after RMA. However, I'm not quite sure what a more appropriate
distribution
would be.
Johannes
Quoting Arne.Muller@aventis.com:
> Sorry, my last posting was incomplete (slipped over the keyboard
...).
>
> I meant that I haven't explored other methods yet, but since the RMA
> values
> are log2, I thought that I'd get something close to a normal
> distribution.
> Comapred to a normal distribution I get many low intensity probe
sets.
>
> The values are generated like this:
>
> eset.rma <- expresso(cel, bgcorrect.method="rma",
> normalize.method="quantiles", pmcorrect.method="pmonly",
> summary.method="medianpolish")
>
> then:
> hist(exprs(eset.rma[,10]))
>
> kind regards,
>
> Arne
>
> --
> Arne Muller, Ph.D.
> Toxicogenomics, Aventis Pharma
> arne dot muller domain=aventis com
>
> _______________________________________________
> Bioconductor mailing list
> Bioconductor@stat.math.ethz.ch
> https://www.stat.math.ethz.ch/mailman/listinfo/bioconductor
>
from the data i have looked at they are in general not normal. i would
not expect them to be.
there are various reasons to take the log. but to make the
within chip distibution normal is not one of them.
what is close to normal is the distribution of RMA
expression measures of a particular gene across many chips. in your
example:
hist(exprs(eset.rma[10,])
but you'll need 30 or more arrays to see it.
On Wed, 3 Mar 2004 Arne.Muller@aventis.com wrote:
> Sorry, my last posting was incomplete (slipped over the keyboard
...).
>
> I meant that I haven't explored other methods yet, but since the RMA
values
> are log2, I thought that I'd get something close to a normal
distribution.
> Comapred to a normal distribution I get many low intensity probe
sets.
>
> The values are generated like this:
>
> eset.rma <- expresso(cel, bgcorrect.method="rma",
> normalize.method="quantiles", pmcorrect.method="pmonly",
> summary.method="medianpolish")
>
> then:
> hist(exprs(eset.rma[,10]))
>
> kind regards,
>
> Arne
>
> --
> Arne Muller, Ph.D.
> Toxicogenomics, Aventis Pharma
> arne dot muller domain=aventis com
>
> _______________________________________________
> Bioconductor mailing list
> Bioconductor@stat.math.ethz.ch
> https://www.stat.math.ethz.ch/mailman/listinfo/bioconductor
>
Anne,
There is no reason why the distribution of expression values across a
collection of genes should be normal. A given gene may have a normal
distribution across samples depending on the selection of samples. If
there
is structure in the sample set (treatment groups, etc), then normality
of
errors around the main effects may be a reasonable assumption. If
your
analysis assumes normality for the distribution of expression values
for a
given sample across all genes, you may want to compare your results
with
those obtained from an analysis that doesn't make this assumption.
Models are fitted to background corrected, normalized probe intensity
data
for each probe set separately. At that level, the distribution of
residuals
is not inconsistent with a contaminated normal error distribution for
which
a robust estimation procedure as used in RMA makes sense.
-francois
----- Original Message -----
From: <arne.muller@aventis.com>
To: <bioconductor@stat.math.ethz.ch>
Sent: Wednesday, March 03, 2004 4:16 AM
Subject: [BioC] RMA and quantile normalisation
> Sorry, my last posting was incomplete (slipped over the keyboard
...).
>
> I meant that I haven't explored other methods yet, but since the RMA
values
> are log2, I thought that I'd get something close to a normal
distribution.
> Comapred to a normal distribution I get many low intensity probe
sets.
>
> The values are generated like this:
>
> eset.rma <- expresso(cel, bgcorrect.method="rma",
> normalize.method="quantiles", pmcorrect.method="pmonly",
> summary.method="medianpolish")
>
> then:
> hist(exprs(eset.rma[,10]))
>
> kind regards,
>
> Arne
>
> --
> Arne Muller, Ph.D.
> Toxicogenomics, Aventis Pharma
> arne dot muller domain=aventis com
>
> _______________________________________________
> Bioconductor mailing list
> Bioconductor@stat.math.ethz.ch
> https://www.stat.math.ethz.ch/mailman/listinfo/bioconductor
>