I have several sets of Illumina data that assigning quality weights is appropriate for a limma analysis. While `arrayWeights` is very frequently used, I have noticed limma has the function `beadCountWeights` that appears to be performing quality weighting for BeadChips, but that function was never discussed here.

So, may I ask: For Illumina chips that have bead data (i.e. Avg_NBEADS and BEAD_STDEV) that can be extracted from GenomeStudio, which is the better way to calculate quality weights?

We have not been able to demonstrate any clear improvement in DE assessment after running beadCountWeights() so, while it is still provided as an option in the limma package, I don't promote or recommend it. It doesn't do any harm either if you do run it.

On the other hand, arrayWeights() is a published method with clear benefits in many circumstances.

The two functions are not equivalent in scope or intention. beadCountWeights() is intended to respond to purely technical quality issues within a beadchip, especially the number of beads for a particular probe. arrayWeights() is much more general and accounts for both technical and biological variability. The latter is often more important.