[BioC] Question about how the lmFit function in Limma uses spot weights

[Jeremy Darot]

Hi,

I'm using Limma to find differentially expressed genes in a set of 2
replicate two-colour arrays, and I would like to know precisely how
lmFit takes into account the weights I gave to the spots.

My question arises because in the topTable of differentially expressed
genes, I see genes for which one spot had been flagged out as bad on
one array, but not on the other. Those genes don't appear in the top
100, so it's not a huge deal, but I was surprised those genes showed
up at all.

Does lmFit give some kind of penalty during the fitting to genes for
which we have fewer data points than the total number of arrays,
because they have been flagged out on some of the arrays?

Cheers,
Jeremy