How to get the different ASVs from the same cluster OTU

[nmgduan]

Hello,

I have two questions.

The first one:

The pair-end reads were used to generate ASVs through DADA2, and then the generated ASVs were clustered into OTUs at 97% identity through qiime vsearch cluster-features-de-novo. Is it feasible?

The second one:

In some articles, the different ASVs from the same clustered OTU were summaried. So how to get the different ASVs from the same clustered OTU?

Thanks!

Mort

[colinbrislawn (Colin J Brislawn)]

Good morning Mort,

@nmgduan

The pair-end reads were used to generate ASVs through DADA2, and then the generated ASVs were clustered into OTUs at 97% identity through qiime vsearch cluster-features-de-novo . Is it feasible?

Yes! You might find that you would just like to use ASVs, and don't have the need for old-school features like OTUs. But it's always nice to compare the old and the new.

@nmgduan

In some articles, the different ASVs from the same clustered OTU were summarized. So how to get the different ASVs from the same clustered OTU?

Not sure the best way to do that... You could make this from the vsearch outputs, but I'm not sure if these are returned by default in the Qiime 2 plugin.

Colin

[Nicholas_Bokulich]

@nmgduan

In some articles, the different ASVs from the same clustered OTU were summaried. So how to get the different ASVs from the same clustered OTU?

They are not returned by default from QIIME 2's q2-vsearch plugin, so you would need to do this in VSEARCH (which fortunately comes installed with QIIME 2!)

[nmgduan]

Thanks for your replies!

But I still have something confused. The pair-end reads were processed using qiime dada2 denoise-paired to get table.qza and rep-seqs.qza. Then the rep-seqs.qza and rep-seqs.qza were use to get OTUs using qiime vsearch cluster-features-de-novo. And you also think it is feasible.
However, when I looked back on this section OTU Clustering, I found four requirements on sequences, one of which is reads are all trimmed to the same length. As far as I can see, the lengths of the sequences in rep-seqs.qza, which is from DADA2, are not the same, so the differences in the sequence lengths will have negative effects on clustering results?

By the way, in this link removing-non-biological-sequences, I think the sentence If you’re going to use DADA2 to denoise your sequences, you can remove **biological** sequences at the same time as you call the denoising function. should be revised to If you’re going to use DADA2 to denoise your sequences, you can remove **non-biological** sequences at the same time as you call the denoising function.

Thanks!

Mort

[colinbrislawn (Colin J Brislawn)]

@nmgduan

Then the rep-seqs.qza and rep-seqs.qza were use to get OTUs using qiime vsearch cluster-features-de-novo . And you also think it is feasible.

Yep, this should work.

@nmgduan

are all trimmed to the same length.

More accurately, trimmed to the same region. A few base pairs of difference is not a problem as long as they are all V4, but you can't cluster reads from V4 and V3-V4 together.

If you run vsearch directly, you can pass the --uc output which will give you each cluster and each hit to a cluster (those hits are different AVS that are all members of that OTU).

Let me know if you need a hand crafting the vsearch command!

Colin

[nmgduan]

Thank you for your prompt reply.

I will have a try. If I have other questions, I will post here!

Thanks!

Mort