which file should I use for generating beta-diversity?

General Discussion
1

I am not sure which table to use for generating the beta-diversity plot. I went through the qiime2-dada2 pipeline and generated two asv tables.
The first one is the table.qza file right after the dada2 step.
The second one is one of the tables I generated after I ran the " qiime taxa collapse" command. This the for loop code I used:
for i in {1..7}

do

qiime taxa collapse \

--i-table table.qza \

--i-taxonomy taxonomy.qza \

--p-level $i \

--o-collapsed-table table-L$i.qza

From my understanding, the first table doesn't have the OTU ID assigned yet.

image
image825×309 12.8 KB

The second table did have the OTU ID assigned. (I used the table at the species level)

image
image1215×266 10.2 KB

This two tables generated two different beta-diversity plots, when I use the R package-vegan.

My question is which one I should use for generating Beta-diversity plot?

Thank you so much!

1 Like
2

Hello again,

I would recommend using the full table made by DADA2, as that includes all ASVs and no database bias.

Once you collapse by taxa level, you group ASVs together based on what the taxonomy the database can resolve, and add database bias: 100s of ASVs may get grouped into a single taxon like p__Bacillota, c__Bacillio,__Lactobacillales; f__? and all this variation at the ASV level is lost.

A summarized taxonomy table that only goes to order or family can be helpful for making bar plots, but I would use the :sparkles: original, maximum resolution ASVs :sparkles: for my beta-diversity calculations and PCoA plots.

P.S.

That table does have a feature ID (in the first column called "#OTU ID") and taxonomy if you have run a classifier, but the full ASV IDs are shown.

2 Likes
3

Thank you so much for answering my question. This is very helpful

1 Like
4

An off-topic reply has been split into a new topic: Using FeatureTable [RelativeFrequency] for Indval Test.

Please keep replies on-topic in the future.

5

This is a great question and I'm curious - if these are ASV IDs, why are they labelled as OTU ID? Is this merely a qiime2 artefact?

6

Different programs label them differently. And ASVs replaced OTUs

Outside the field of amplicons, we might more generally call ASVs / OTUs 'features'. This is the same as a feature abundance table or an abundance matrix in numerical ecology.

7

Hi Colin, yes familiar with the paper and why ASVs have replaced OTUs/ feature usage. I was curious as to why qiime2 output still says OTU ID when fed in ASV table input.

1 Like
8

Want to post the output so we can take a look? Maybe we should update our labels :writing_hand:

1 Like