[BioC] questions on BioC's implementation of plier

[Jenny Drnevich]

Hi Crispin & others,

I was comparing the plier implementation you've done with Affymetrix's 
technical note on plier 
(http://www.affymetrix.com/support/technical/technotes/plier_technote.pdf) 
and some of the options don't seem to match. In Affy's technical note, 
after the normalization option there is a background step, with the options 
'PM-MM', 'PM-B', 'PM only', and 'MM as additional PM'. However, in BioC's 
justPlier(), after the 'normalize' argument there is a 'norm.type' argument 
with options 'separate', 'pmonly', 'mmonly' or 'together'.  Looking at the 
code, I see 'norm.type' gets pushed to normalize.AffyBatch.quantiles(). I 
take it 'norm.type' is just additional functionality added to the 
normalization step?

How are the background options implemented? There are 'usemm' and 
'usemodel' arguments in justPlier(), but there's no information given about 
them, other than the defaults are usemm=TRUE and usemodel=FALSE. I did 
create my own wrapper around justPlier to add the variance stabilization at 
the end (below). Because Affy's presentation of plier highlights the 
background correction options, it would be better if justPlier() had a more 
intuitive interface to them, with a better explanation in the help page. 
Adding the variance stabilization option would also enhance the 
user-friendliness of justPlier.

Thanks,
Jenny

JDplier <- function (object, var.stab=T, stab.val=16, ... )
{
     object <- justPlier(object, ...)

     if (var.stab)  exprs(object) <- log2(2 ^ exprs(object) + stab.val)

     return(object)
}


Jenny Drnevich, Ph.D.

Functional Genomics Bioinformatics Specialist
W.M. Keck Center for Comparative and Functional Genomics
Roy J. Carver Biotechnology Center
University of Illinois, Urbana-Champaign

330 ERML
1201 W. Gregory Dr.
Urbana, IL 61801
USA

ph: 217-244-7355
fax: 217-265-5066
e-mail: drnevich at uiuc.edu