[BioC] Basic question about Limma

[Heidi Dvinge]

On 28 Apr 2010, at 11:11, Guido Leoni wrote:

> Thank you Heidi and Neel
> I'll give you more details about my data:
> The experiments are performed with an affymetrix custom 3' array
> here is an example for one probeset :
> Normalized expression with gcRMA
>                                               control        
> control        control       case          case           case
> <!-- BODY,DIV,TABLE,THEAD,TBODY,TFOOT,TR,TH,TD,P { font- 
> family:"Arial"; font-size:x-small } -->
> 219070_s_at 679,15 734,98 777,32 667,32 507,76 645,37
>
So, you data is definitely not on a log2 scale, since it'd then be  
from 0-16 (rather than 2^0-2^16). limma expects the MAList (or the  
other inputs to lmFit) to contain log2 values, c.f. the documentation  
for ?lmFit

     lmFit 
(object,design=NULL,ndups=1,spacing=1,block=NULL,correlation,weights=NUL 
L,method="ls",...)

Arguments:

   object: object of class ‘numeric’, ‘matrix’, ‘MAList’, ‘EList’,
           ‘marrayNorm’, ‘ExpressionSet’ or ‘PLMset’ containing
           log-ratios or log-values of expression for a series of
           microarrays

Without knowing exactly what you did, i.e. seeing your actual code,  
I'm afraid this list can't give you any more specific help. But this  
is the source of the unexpected behaviour you see for your log2 fold  
changes.

HTH
\Heidi



> This is the corrisponden row from limma output  for the same probeset
> AFFy ID                 EG        Symbol  M                        
> A                        t                           
> P.value                    B  <!--  
> BODY,DIV,TABLE,THEAD,TBODY,TFOOT,TR,TH,TD,P { font-family:"Arial";  
> font-size:x-small } --> <!--  
> BODY,DIV,TABLE,THEAD,TBODY,TFOOT,TR,TH,TD,P { font-family:"Arial";  
> font-size:x-small } -->
> 219070_s_at 64598        MOSPD3 -124.008563236060 666.759701175327  
> -2.27688122341071 0.0489596557364987 -4.19921794672031
>
>
>
> 2010/4/27 Heidi Dvinge <heidi at ebi.ac.uk>
> Hello Guido,
>
> the M-values are indeed the log2 fold change, so log2(R/G) for the  
> red and
> green channel. However, after normalisation the R and G intensities  
> are
> usually reported on a log2 scale (from 0-16), and since
> log2(R/G)=log2(R)-log2(G) that's probably what you see.
>
> HTH
> \Heidi
>
> > Dear List I've a very basic question about the  limma output (and  
> I'm
> > newbie
> > of MA analysis).
> > I'm looking some analysis output .Usually in the M column (if I well
> > understand ) I should expect to find the log2-fold change of my
> > conditions.Instead I find the difference between the normalized
> > expressions...
> > Some one could suggest to me how to transform these values in  
> log2  fold
> > change?
> > Thank you very much and sorry for my silly question
> >  Guido Leoni
> >
> >       [[alternative HTML version deleted]]
> >
> > _______________________________________________
> > 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
> >
>
>
>
>
>
> -- 
> Guido Leoni
> National Research Institute on Food and Nutrition
> (I.N.R.A.N.)
> via Ardeatina 546
> 00178 Rome
> Italy
>
> tel     + 39 06 51 49 41 (operator)
>        + 39 06 51 49 4519 (direct)