[BioC] Use of individual channel from a 2-color array as a measure of expression level?

J.delasHeras at ed.ac.uk J.delasHeras at ed.ac.uk
Thu Jun 9 13:59:32 CEST 2011

Quoting Hari Easwaran <hariharan.pe at gmail.com> on Wed, 8 Jun 2011  
12:42:12 -0400:

> Dear BioC gurus,
> Two-color arrays are generally used to compare expression of a test sample
> (like drug treated) relative to control. Usually Cy5 channel is used for the
> test and Cy3 is used for the control. The output from such data that is used
> for any analysis is the log ratios that gives an idea of relative expression
> of genes in the test vs control samples.
> Can the intensities from an individual channel (Cy3) in a two-color
> microarray expression data (in my case Agilent) be used to get a measure of:
> (a) absolute (approximate) expression level (say E) of a gene/probe
> (b) can the Es for a gene/probe from two or more control samples be compared
> to get an idea of the general expression of that gene in the samples. If
> yes, what is the best way to normalize between the Cy3 channels from
> different arrays?
> I guess my question is can the individual channels from a two-color array be
> used like a one-color array.
> Sorry if such an issue has been addressed before, but I couldn't get
> anything on the web (the keywords to search on such a topic is damn vague).
> I would appreciate any help/suggestions.
> Sincerely,
> Hari

Dear Hari,

there is a way to treat each of a two-colour hyb (no particular  
convention as to which one is Cy3 and which one is Cy5, in fact many  
still include dye swap experiments) as a single hyb for various  
analysis. The Limma user's guide has a chapter on it and it's a good  
place to start.

However, if what you want is to compare relative expression levels  
between probes/genes to establish that transcript A is expressed 2x  
the level of transcript B, I don't think you can do that.

The reason is that different probes seem to have a different  
background level and a different intensity range. This is illustrated  
very clearly if you check the individual probes in a probeset that  
match a single transcript. Sometimes they all are roughly at the same  
level, but very often you get some probes that are very low or very  
high. Very often.
For that reason, it is safe to compare the behaviour of the same probe  
in a range of conditions/samples, but not to compare between probes.
You could compare between probes, I suppose, if you use an  
intermediate comparison, some kind of "normaliser". But I suspect it's  
not a trivial task to get any kind of trustworthy results.

Microarray analyses are best used to compare (many) individual  
probes/probesets across various conditions or samples, whether using a  
single channel or a two channel approach, it doesn't really change  
matters, and even that is not very quantitative unless you include and  
use standards in your array design & hybridisation.


Dr. Jose I. de las Heras                      Email: J.delasHeras at ed.ac.uk
The Wellcome Trust Centre for Cell Biology    Phone: +44 (0)131 6507090
Institute for Cell & Molecular Biology        Fax:   +44 (0)131 6507360
Swann Building, Mayfield Road
University of Edinburgh
Edinburgh EH9 3JR

The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.

More information about the Bioconductor mailing list