High similarity (>97%) in ASV table

I have created an OTU table (ASV) using DADA2 script.
When I analyzed the different OTUs according to their sequences, I have noticed that some of them are resembled to each other (according to their sequences) in more than 97% and even more than 99% identity by using blast.
My question is why?
I went to get an OTU table which every OTU (ASV) is a cluster of 97% (identity of sequences).
I looked at the script again and I didn’t see any parameter which allows to set the identity level in each OTU.
Am I missing something here?
When I used the assign taxonomy script afterward, I worked with the 97% db.
Tx a lot,

1 Like

Hello Nir,

Thank you for posting on the forums.


OTUs and ASVs are different, and it sounds like you were expecting OTUs, and were surprised to get ASVs.
https://www.nature.com/articles/ismej2017119 :point_left:

DADA2 makes ASVs, which can be different by only a single base pair! Your feature table contains ASVs, which is why some of them are very similar.

Let me know if that helps answer your question!



Hi! If even after @colinbrislawn clarification you still want to work with OTUs instead of ASVs, you can cluster them at 97% of identity https://docs.qiime2.org/2019.10/tutorials/otu-clustering/.
But ASVs are allowing to perform your analysis with higher resolution, IMHO.

1 Like

Hi Colin,
Thanks for your answer, I will read the reference to better understand this process.

1 Like

I prefer to stay with ASV, as some of my analyses are related to species and strain levels.
I didn’t realize that different clusters (ASV) could be similar to each other by even a single bp.
Thanks again,


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