gneiss
gneiss copied to clipboard
Median and dispersion estimators
This is building off of the differential abundance estimators used in ALDEx2
This may help with accounting for pseudocounts.
Basically, this would bootstrap a Dirichlet prior, and take the median and dispersion as a proxy for the mean and variance for the estimator.
Ultimately, this will allow us to account for variances in the low counts.
CC @ggloor