Score = 0 for all tests and elaboration on running PCoA prior to DAtest about datest

[Russel88]

Hi Carly,

That all methods have a Score <= 0 does not mean that no features differ. It means that all methods have a false discovery rate as least as high the emperical power (adjusted by the AUC, see wiki for details).

DAtest uses the variation in your dataset to estimate how well different methods perform. But if much of this variation comes from the predictor of interest the methods will appear worse than they actually are.

This does not mean that you cannot use tests of differential abundance, you just cannot rely on the ranking of methods that DAtest gives you. You can however use the False Positive Rate to filter out methods that give too many false positives (this does not rely on spike-ins, and is therefore not affected by this issue).

Hope this makes sense,

Cheers,