[BioC] 'A' mean in lmFit function in limma?

[Gordon K Smyth]

The Amean values are intended to give an idea of the overall expression
level for each gene.  The A-values are not used to analyze differential
expression and therefore there is usually no purpose is fitting a linear
model to them, which is what would give you the same output format as the
M-values.

If you want to see the mean A-values corresponding to the M-values for
each group, you could simply use

fitA <- lmFit(MA$A, design)

and look in fitA$coef.  Note this only works when the design matrix is a
simple indicator matrix for the RNA groups as in your example.  There are
no plans to specially support this sort of analysis in limma.

Gordon

> Im trying to find the average A values for replicate chips after using
> the lmFit function. I can easily identify the average M values for each
> group of replicates, but I dont see a similar listing for the A values.
> I only appear to be generating one average A value for all replicates.
> Does the lmFit function do this, or am I doing something wrong...
>
> thanks
> Simon.
...
> $Amean #why isnt the format the same as for M?, there only appears to
> be one set of A values#
> [1] 10.967244 15.627646 15.384239  7.634115 12.521264
> 25083 more elements ...