# classify.otu algoritm

**URL:** <https://forum.mothur.org/t/classify-otu-algoritm/906>\
**Category:** Commands in mothur\
**Created:** [May 16, 2012, 1:27pm UTC](https://forum.mothur.org/t/classify-otu-algoritm/906 "2012-05-16T13:27:18Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![Anna\_B](https://avatars.discourse-cdn.com/v4/letter/a/59ef9b/32.png) [@Anna\_B](https://forum.mothur.org/u/Anna_B)\
**Post date:** [May 16, 2012, 1:27pm UTC](https://forum.mothur.org/t/classify-otu-algoritm/906/1 "2012-05-16T13:27:18Z")

</div>

Hi everyone, I have a doubt regarding classify.otus and cutoffs. I get different results doing:

1. classify.seqs without specifying cutoff and then calcualte OTus and classify.otus using a 80% boostrap
2. classify.seqs specifying a 80% boostrap and then calcualte OTus and classify.otus using a 80% boostrap  
The result is that I get a lot more classified OTUs with the first option, but I don’t know how classify.otus calculate the consensus taxonomy for the taxon, so I don’t know if I can trust. What do you recommend?

---

<div class="post-metadata">

**Author:** ![pschloss](https://yyz2.discourse-cdn.com/flex036/user_avatar/forum.mothur.org/pschloss/32/4_2.png) [@pschloss](https://forum.mothur.org/u/pschloss)\
**Post date:** [May 16, 2012, 6:24pm UTC](https://forum.mothur.org/t/classify-otu-algoritm/906/2 "2012-05-16T18:24:29Z")

</div>

The 80% in classify.seqs is the confidence score for the classification of that sequence

The 80% in classify.otu means that 80% of the reads in an OTU have that classification.

Subtle, but different. We recommend 80% in classify.seqs and the default (50%) in classify.otu.

Pat

---

<div class="post-metadata">

**Author:** ![Anna\_B](https://avatars.discourse-cdn.com/v4/letter/a/59ef9b/32.png) [@Anna\_B](https://forum.mothur.org/u/Anna_B)\
**Post date:** [May 17, 2012, 6:46am UTC](https://forum.mothur.org/t/classify-otu-algoritm/906/3 "2012-05-17T06:46:39Z")

</div>

Thanks a lot!!

---

<div class="post-metadata">

**Author:** ![Shaunson26](https://avatars.discourse-cdn.com/v4/letter/s/9fc29f/32.png) [@Shaunson26](https://forum.mothur.org/u/Shaunson26)\
**Post date:** [April 18, 2013, 7:31am UTC](https://forum.mothur.org/t/classify-otu-algoritm/906/4 "2013-04-18T07:31:27Z")

</div>

Is there a better way to classify OTUs? How does the classification of an OTU work if there is GroupA(100) and GroupA(99) in the cluster? Can there be a weight added to the final ‘confidence level’ of an OTU rather than how many were the same? e.g GroupA(100) and GroupA(99) = GroupA(99.5)