The place of feature rarefaction when using `picrust2`

picrust / picrust2

Code, unit tests, and tutorials for running PICRUSt2

GNU General Public License v3.0

329 stars 104 forks source link

The place of feature rarefaction when using `picrust2` #201

Closed valentynbez closed 2 years ago

valentynbez commented 3 years ago

Hi,

this is more of a general methodological question. If we use PICRUSt2, when in the pipeline should be samples rarefied? Is it on the OTU-table, before plugging it in? Or after predictions are made, on feature tables? Which would be the best practice?

Cheers

gavinmdouglas commented 3 years ago

Hi @crusher083, this question is related to this FAQ: https://github.com/picrust/picrust2/wiki/Frequently-Asked-Questions#should-i-rarefy-or-transform-the-predictions-after-running-picrust2

I don't rarefy microbiome count data in general - for PICRUSt2 I would not rarefy the data at any point and either convert to relative abundances at the end or use a tool that can handle compositional data.