Dear Adrien,
> I'm trying to integrate the results of two experiments, one
involving
> moe430 A and B chips, the other moe430_2 chips.
> Hence I thought about combining AffyBatches from moe430 A and B into
> one, to get moe430_2 "equivalents".
>
> This is relatively easy to do after processing (i.e. at the probeset
> level), but the trick is that I would really like to do this at the
> probe level (because I would like to normalize the combined chips
> together with moe430_2 chips - not sure if this make sense by the
way).
It might indeed not make sense, if you use standard normalization
methods, since in general you need a separate normalization
transformation for each array.
What I would try is to set up big linear model with appropriate array
effects (e.g. one background and one scale parameter for each array),
and come up with an efficient enough method for fitting it. I don't
know
of any existing software for this - perhaps others do?
For the actual stitching the data together, you might want to check
the
code in the combineAffyBatch function in the matchprobes package;
while
it does not exactly do what you want, maybe some of it is reusable.
> 1) Does anyone know if there is a function written somewhere that
could
> help?
>
> 2) Or, would it be possible to create a moe430_2 AffyBatch object
from
> scratch?
> (I could probably read an existing chip and replace the PMs and MMs
but
> that's not very elegant, is it?)
>
> 3) Or, if nothing else, would there be a function to do quantile
> normalization at the probeset level?
>
> Thanks for any ideas...
> Adrien
Best regards
Wolfgang
-------------------------------------
Wolfgang Huber
European Bioinformatics Institute
European Molecular Biology Laboratory
Cambridge CB10 1SD
England
Phone: +44 1223 494642
Fax: +44 1223 494486
Http: www.ebi.ac.uk/huber
Agreed. Also see the following thread
http://files.protsuggest.org/biocond/html/3268.html
Regards, Adai
On Fri, 2005-07-22 at 11:34 +0100, Wolfgang Huber wrote:
> Dear Adrien,
>
> > I'm trying to integrate the results of two experiments, one
involving
> > moe430 A and B chips, the other moe430_2 chips.
> > Hence I thought about combining AffyBatches from moe430 A and B
into
> > one, to get moe430_2 "equivalents".
> >
> > This is relatively easy to do after processing (i.e. at the
probeset
> > level), but the trick is that I would really like to do this at
the
> > probe level (because I would like to normalize the combined chips
> > together with moe430_2 chips - not sure if this make sense by the
way).
>
> It might indeed not make sense, if you use standard normalization
> methods, since in general you need a separate normalization
> transformation for each array.
>
> What I would try is to set up big linear model with appropriate
array
> effects (e.g. one background and one scale parameter for each
array),
> and come up with an efficient enough method for fitting it. I don't
know
> of any existing software for this - perhaps others do?
>
> For the actual stitching the data together, you might want to check
the
> code in the combineAffyBatch function in the matchprobes package;
while
> it does not exactly do what you want, maybe some of it is reusable.
>
> > 1) Does anyone know if there is a function written somewhere that
could
> > help?
> >
> > 2) Or, would it be possible to create a moe430_2 AffyBatch object
from
> > scratch?
> > (I could probably read an existing chip and replace the PMs and
MMs but
> > that's not very elegant, is it?)
> >
> > 3) Or, if nothing else, would there be a function to do quantile
> > normalization at the probeset level?
> >
> > Thanks for any ideas...
> > Adrien
>
> Best regards
> Wolfgang
>
> -------------------------------------
> Wolfgang Huber
> European Bioinformatics Institute
> European Molecular Biology Laboratory
> Cambridge CB10 1SD
> England
> Phone: +44 1223 494642
> Fax: +44 1223 494486
> Http: www.ebi.ac.uk/huber
>
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/bioconductor
>
