I tried searching for this elsewhere but could not find it on the forums, I apologize if I missed it. What is the data transformation (square root, log, etc…) used in the core-metrics-phylogenetic script on the counts data? What is your general thoughts regarding which transformation (square root, some version of log) to use before say Bray-Curtis and/or Unifrac distance based analysis and subsequent PCoA analysis?
The data is rarefied to your specified depth (a subsampling without replacement and current necessary evil) and then diversity is calculated. If you drop your artifact into qiime2 view and click on the provenance tab, you will find all the commands used to generate that artifact.
Thank you for your quick response! I took a look at the provenence and it just states that the core_metrics_phylogenetic script was run at the last step for all the qzas and qzvs generated by that step.
Sorry just to clarify, based on what you are saying, and from what I can tell from the script on github, there is no other data transformation besides rarefying used to generate the diversity metrics? No additional square root or log transformation? Just making sure I understand for writing up my methods if I were to use one of the visualizations.
Sorry for the mix up! I guess I dont know my pipeline provances as well as I should. (Although it's still a QIIME trick I appreciate for my methods section. If there is a specific transform, it would be wrapped into the metric calculation. My methods sections for distance often read something like,
Data was rarefied to [depth] sequences/sample. [metric] distances were calculated on the rarefied data (cite specific metrics here). PCoA projections were visualized using Emperor (cite Emperor and/or animated Emperor here).
(PS the "citations" for the specific things can be found under the citations tab)