[BioC] One directional spatial normalisation for microarray data

[Henrik Bengtsson]

Hi,

just to make sure - the systematic effects you have, is it possible
that they are due to variability between microtiter plates?  Since you
say you only have spatial effects in one direction, this could the
case.  If so, when spotting arrays from such plates row by row (or
column by column), in an NxN grid of print tips, each print-tip
group/grid will show horizontal (or vertical) bands.  See

H. Bengtsson, Identification and normalization of plate effect in cDNA
microarray data, Preprints in Mathematical Sciences, 2002:28,
Mathematical Statistics, Lund University, 2002.
URL: http://www.maths.lth.se/bioinformatics/publications/

Do you have bands like in left panel of Figure 1.  If you plot the
data in print order, you might then get what is in Figure 2.  If so,
it is very likely there is print-order (print-dip) effect in you data.

*HOWEVER*, this does *not* mean that you should do print-order
normalization (early on I did said you should do this, but I know
better know).  Instead, think about it this way.  If different plates
gives different probe *intensities*, and this is expected, an
intensity-dependent artifact in log-ratios will the show up as a
printorder effect ("symptom") on log-ratios.  However, if you remove
intensity-dependent artifacts in you log-ratios, the printorder
symptoms should go away.  If it doesn't, you can probably to a better
job in removing intensity-dependent artifacts.   I don't have final
documentation on this, but I think the following talk will illustrate
it all:

H. Bengtsson, Plate Effects in cDNA Microarray Data, Statistics
Department, University of California in Berkeley. September 26, 2002.
URL: http://www.maths.lth.se/bioinformatics/calendar/20020926/

In the aroma package (http://www.braju.com/R/) there are several
different plot functions that allows you to plot your data in
printtip/-dip order.

Hope this helps

Henrik


On 26/11/2007, michael watson (IAH-C) <michael.watson at bbsrc.ac.uk> wrote:
> Hi
>
> I'm looking for a normalisation procedure which will normalise log ratio
> based on row only.  I know the marray package has 2D normalisation, but
> I fired that at my data and it didn't touch it - the data I have is
> independent of column, but very dependent on row.
>
> Is there anything?
>
> Thanks
> Mick
>
> The information contained in this message may be confide...{{dropped:14}}
>
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/bioconductor
> Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor
>