Alpha diversity with filtered raw data or unfiltered raw data

General Discussion
, , ,
1

Hi,

I'm working on my qiime2 output in R and I filtered for low abundance taxa and low reads in my phyloseq object. I'm wondering if I should estimate alpha diversity on the unfiltered untrimmed phyloseq object or the trimmed one.

I read some papers and previous discussions on qiime2 forum, most recommend against observation alpha diversity metrics since it relies on singletons which are removed during DADA2 workflow. I'm mostly interested in Berger-Parker, Shannon, Simpson, and Faith.

I have decided not to proceed with any normalization for all of them except berger-parker since it measures the relative abundance anyway, so I normalized by relative abundance for that one metric, the rest has raw data.

Thank you :smile:

1 Like
2

Hello Saif!

I'm going to to leave this here: Rarefaction is currently the best approach to control for uneven sequencing effort in amplicon sequence analyses - PMC

Here's a related Qiime2 plugin: Introduction 🥾 - q2-boots Documentation

2 Likes