[BioC] M vs A plots of affymetrix data

[michael watson (IAH-C)]

Hi Ben
 
Thanks for the help :)
 
When I tried following your example on the affybatch.example data set:
 
eset.rma <- rma(affybatch.example)
MAplot(eset.rma)
 
I get an error about there being no direct or inherited method.
 
Mick

________________________________

From: Ben Bolstad [mailto:bmb at bmbolstad.com]
Sent: Thu 09/03/2006 3:15 PM
To: michael watson (IAH-C)
Cc: bioconductor at stat.math.ethz.ch
Subject: Re: [BioC] M vs A plots of affymetrix data




> I realise that I have to take two arrays pairwise to get my ratio(M)
> and average(A), but does anyone know if these plots are generally at
> the probe level or at the gene level?

You can do them at the probe or probeset level. In the document you
referenced some were done at the probe level and some were done at the
probeset level.


> If at the probe level, are they the ratio and average of just the PM
> intensities?

Typically just PM intensities.


> Also, when I read data in using ReadAffy(), are the PM and MM
> intensities raw or on a logged scale?  If I then perform rma() or
> mas5(), are the gene level intensities on a log scale?  This is
> important, as I need to figure out how to calculate the ratio and
> average

ReadAffy() returns intensities on the natural scale. rma() expression
values are on the log2 scale. mas5() values are natural scale.

You of course are free to construct these yourself, but there are some
buit-in functions for doint this also. If you have affyPLM loaded you
should be able to use the MAplot() function applied to either
AffyBatches or exprSets.  eg


library(affyPLM)
data(Dilution) # Dilution is an AffyBatch.

# each array against a median-wise synthetic reference array
MAplot(Dilution,ylim=c(-2,2))

# comparing arrays 2:4 with array 1
MAplot(Dilution,ylim=c(-2,2),ref=1,which=2:4)


# MA plot comparing 1 to 4
MAplot(Dilution,ylim=c(-2,2),ref=1,which=4)


#the infamous pairwise plot
MAplot(Dilution,pairs=TRUE)


eset.rma <- rma(Dilution)

#examine to see that they are on log2 scale
exprs(eset.rma)[1:5,] 

# each array against a median-wise synthetic reference array
MAplot(eset.rma)


eset.mas5 <- mas5(Dilution)

#examine to see on natural scale
exprs(eset.mas5)[1:5,]


#need to tell it to take the log if supplying natural scale
#expression values.
MAplot(eset.mas5,log=TRUE)




--
Ben Bolstad <bmb at bmbolstad.com>
http://bmbolstad.com