Rarefaction and alpha diversity

Hi @gregcaporaso,

I am confused about the rarefaction. Why do we need to do the rarefaction? What is the benefit? In addition, if we do subsampling, is there a command line to repeat the subsampling results? Or we just need to check the subsampling results form the rarefied_table.qza? Besides, when I run this qiime diversity core-metrics command, chao1 diversity measurement is not included in the output. If I want to get the chao1 measurement, how can I get it? Is the qiime diversity alpha command based on the subsampled data?
Thank you so much!

1 Like

Hi @pumpkin,

Even subsampling is necessary for alpha diversity estimation because we need to control for uneven sampling. For example, imagine that we have 2 different soil samples and want to determine which one contains the most unique species. Sample 1 has 1000 sequence reads, representing a total of 100 unique species; sample 2 has 100 sequence reads representing a total of 50 unique species. Which one has more unique species? Without rarefying, sample 1 has more species, but after evenly rarefying sample 2 would have more.

Check out feature-table rarefy

Sounds like you already found the way: alpha

You should provide alpha with a rarefied feature table. Running feature-table rarefy, then alpha with the rarefied table would replicate the alpha diversity results output from core-metrics.

I hope that helps!


Hi @Nicholas_Bokulich,

Thanks for your quick response. You answer really make sense.


@pumpkin, a quick follow-up on @Nicholas_Bokulich’s answer, as of QIIME 2 2017.9 both core-metrics and core-metrics-phylogenetic now return the rarefied feature table, so if you want to use the same rarefied table that was used to compute the values returned in core-metrics/core-metrics-phylogenetic, you can just use that artifact (instead of recomputing the rarefied table). Either approach is good, and probably warrants comparison as part of your analysis, I just wanted to let you know there are a few ways to accomplish this goal! Thanks for using QIIME 2! :t_rex:


This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.