[BioC] LIMMA - Intensity normalization methods

[ninoochka at free.fr]

Hi,

i want to work on normalized intensity values in my experiment (oligo arrays).
To do this, I first need to normalize data with a "within array normalization",
followed by a "between array normization".
So, I used printtiploess function followed by the Aquantile function. Just for
testing purpose,  I also tried with with "quantile" method and I was surprised
to see that toptables were really different: resulting p-values were better with
this second method. And, if I take the first 100 genes in the toptables, 50% are
different. As quantile function normalizes again ratios, does it mean that
printiploess method did not correctly normalized my ratios ?

Here are my first genes:

PrinTipLoess + Quantile:


        "logFC" "AveExpr" "P.Value"
"8802"         2,44         10,86         4,20E-17
"9786"         1,96         10,77         7,51E-17
"8906"         2,24         10,75         7,21E-16
"11808" 2,14         10,74         8,86E-16
"5276"         1,93         10,71         1,55E-15

PrinTipLoess + AQuantile:


        "logFC" "AveExpr" "P.Value"
"8802"         2,73         10,84         1,69E-15
"7678"         1,51         7,96         4,45E-15
"7657"         1,48         8,59         6,59E-15
"9254"         1,71         8,66         8,18E-15
"9786"         2,41         10,75         2,04E-14


Sincerely,

Lucie