[BioC] When to treat technical reps as biological reps? WAS:Re: 2x2 factorial loop without common reference (pool)

[Gordon K Smyth]

On Mon, May 1, 2006 10:16 am I wrote
> BTW, in the limma approach to multilevel models, the full DF is always used.  The DF issues
> associated with ANOVA models is side-stepped.  This is a consequence of the extreme smoothing
> across genes using by the approach.  A full explanation of this would need a lot of space ...

I thought of a way to explain it, which I hope will make some sense.

The limma model to multilevel models (blocking) collapses the two levels of variation down to one
level by imposing a constraint.  The constraint is that the ratio of biological to technical
variation is roughly the same for each gene.  This allows all the arrays, even the technical
replication, to be used in estimating the standard error for contrasts of interest.

Best wishes
Gordon