[BioC] RE : Limma: Normalization with large numbers ofdifferentially expressed genes

J.delasHeras at ed.ac.uk J.delasHeras at ed.ac.uk
Wed Oct 10 16:54:54 CEST 2007


Hi Serge,

> For me it seems as the parameters used for the loess normalization work out
> fine in this situation. Here you may find examples of the MA plots before
> and after normalization:
> http://www.lbmcc.lu/microarray_pics/MA_plots_13.png
> http://www.lbmcc.lu/microarray_pics/MA_plots_15.png
> The two slides shown here are for the timepoint were the most significantly
> regulated genes have been detected.

I suggest you look at the MA plots in a full glorious dolby surround  
sound rather than all squashed up like you have here :-) (make them  
larger, do not compress the A axis, and maybe use smaller dots)
However they do not look like they would be too bad, from these  
pictures. It's hard to get into details with this resolution, but I  
don't see much cause for concern, as you indicate.

> What we intend to do now is to perform Real-Time PCR validation on a larger
> scale testing 40 up to 50 genes over the whole intensity scale. This might
> perhaps help us to see if there were larger problems during loess
> normalization and then integrate this knowledge into normalization.

That sounds reasonable.

> The biology behind the molecule we use in this experiment in relation with
> the used cell line is not so well described in the literature.

Maybe not, but maybe you can make an educated guess as to what to  
expect... perhaps not to suggest a target directly, but to have an  
idea whether you can expect many genes being affected or not, that  
sort of thing...

> But I found
> that for other biological systems used in conjunction with this drug there
> exists a good agreement for the positively and negatively regulated genes
> with our results.

There you go. That sounds again ok.

> The spike-in fold changes in our experiment showed also no
> abnormal behavior compared to the expected values.

It all seems to add up fine... I'd just focus on the RT confirmation  
and if that also is in agreement... it's all okay, I suppose!

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

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



More information about the Bioconductor mailing list