[BioC] Limma: background correction. Use or ignore?

[J.delasHeras at ed.ac.uk]

Quoting Naomi Altman <naomi at stat.psu.edu>:


[Hide Quoted Text]
I have investigated this (somewhat) experimentally.  Background
correction increases the variability of low-expression genes and
reduces it for high expression.  This corresponds to the RMA noise
model since background correction would double the additive variance
but not affect the multiplicative variance (which is the dominant
source of variance for highly expressing genes.)

--Naomi

Thank you Naomi, I think this is one of the main reasons my new arrays 
suffered more
noticeably after bkg-substraction. They contain a large number of low 
and mid-low
intensity spots.

I was even considering making two subsets of my spots (higher and lower 
intensity),
taking those that consistently have lower signal, and analysing them 
separately (in fact,
scanning the slides at higher power, to obtain stronger signals from 
those). I am not
sure if I will be able to get more consistent results that way. I can 
scan at higher
power and get reasonable intensities from the weaker spots, whilest 
keeping a low
background... but then I have a lot of spots that are totally saturated 
and useless...
that's why I thought of the division of the data set and double scan. 
But I haven't tried
it yet.

Jose

-- 
Dr. Jose I. de las Heras                      Email: J.delasHeras at ed.ac.uk
The Wellcome Trust Centre for Cell Biology    Phone: +44 (0)131 6513374
Institute for Cell & Molecular Biology        Fax:   +44 (0)131 6507360
Swann Building, Mayfield Road
University of Edinburgh
Edinburgh EH9 3JR
UK