[BioC] normalizing 133 a and b

Mark Reimers Mark.Reimers at biosci.ki.se
Fri Nov 7 20:59:18 MET 2003

Message: 7
Date: Thu, 6 Nov 2003 12:18:14 +1100
From: Rob Dunne <Rob.Dunne at csiro.au>
Hi Rob,
I think it's unnecessary and even unadvisable to adjust B to the A chips (your 
step 3 & 4). The affy package does a good job of normalizing both. Do you 
find consistent discrepancies between results from A and B chips?


Subject: [BioC] normalizing 133 a and b
To: bioconductor at stat.math.ethz.ch
Message-ID: <16297.41302.579995.483402 at pride.nsw.cmis.CSIRO.AU>
Content-Type: text/plain; charset=us-ascii

HI list,

how are people normalizing 133a and b chips?

here is what I am doing.

1) bg.correct.rma2 (so this is per chip)
2) extract the pm values for all chips
3) for each pair (A and B) get the probes in common (168 genes)
  and make a linear correction factor to to correct the B chip.
4) adjust all the B chips to thir corresponding A chip 
5) put the data (A and B) in a matrix and 
do quantile normalization
6) this gives me the corrected probes, unfortunately some
are now negative so I add a slight fudge factor
7) I read the probes back into an AffyBatch object and get
expression values (medianpolish)

                                        does this sound reasonable?


Rob Dunne         Fax: +61 2 9325 3200     Tel: +61 2 9325 3263
CSIRO Mathematical and Information Sciences     +61 2 9325 3100
Locked Bag 17, North Ryde, New South Wales, Australia, 1670         
http://matilda.vu.edu.au/~dunne  Email: Rob.Dunne at csiro.au

        Java has certainly revolutionized marketing and litigation.

Mark Reimers,
Assistant Professor, Department of Biosciences, and
Statistical Expert, Expression Analysis Facility,
Karolinska Institute
phone: +46-8-608-3333;

More information about the Bioconductor mailing list