# classification **URL:** \ **Category:** Commands in mothur\ **Created:** [May 22, 2017, 9:14am UTC](https://forum.mothur.org/t/classification/3263 "2017-05-22T09:14:25Z")\ **Posts on this page:** 10\ **Page:** 1
**Author:** ![sje062](https://avatars.discourse-cdn.com/v4/letter/s/e8c25b/32.png) [@sje062](https://forum.mothur.org/u/sje062)\ **Post date:** [May 22, 2017, 9:14am UTC](https://forum.mothur.org/t/classification/3263/1 "2017-05-22T09:14:25Z")
Hi mothur forum, may I have a solution, if possible, on how to further group sequences from an OTU that is defined by taxonomy, Bacteria unclassified? I have the rep seq sequence but like to know more about this big group of barely Silva v128 recognised sequences. Sigmund ---
**Author:** ![Kendra](https://yyz2.discourse-cdn.com/flex036/user_avatar/forum.mothur.org/kendra/32/112_2.png) [@Kendra](https://forum.mothur.org/u/Kendra)\ **Post date:** [May 22, 2017, 2:44pm UTC](https://forum.mothur.org/t/classification/3263/2 "2017-05-22T14:44:13Z")
try blasting the rep seq against Ref\_seq ---
**Author:** ![sje062](https://avatars.discourse-cdn.com/v4/letter/s/e8c25b/32.png) [@sje062](https://forum.mothur.org/u/sje062)\ **Post date:** [May 23, 2017, 1:56pm UTC](https://forum.mothur.org/t/classification/3263/3 "2017-05-23T13:56:08Z")
Thanks, I did that (89% closest) and could do it for all but it would have been easier if the ca 6000 sequences could, somehow, have been split into smaller groups so I could blast representatives from each such subcluster. The sequences are from different regions of the 16S so don’t think they align well. The clustering by phylotypes returned that big OTU classified as Bacteria. I suspect this OTU includes a lot of different sequences. Further comments whould be helpful, on how to cluster unaligned unclassified 16S sequences. Sigmund ---
**Author:** ![Kendra](https://yyz2.discourse-cdn.com/flex036/user_avatar/forum.mothur.org/kendra/32/112_2.png) [@Kendra](https://forum.mothur.org/u/Kendra)\ **Post date:** [May 23, 2017, 5:23pm UTC](https://forum.mothur.org/t/classification/3263/4 "2017-05-23T17:23:29Z")
What are your samples? for things like soil, there won’t likely be many closer relatives in ref\_seq. You just have to embrace the unknown If you really want to try, you can cluster the unknown into a higher level OTU (say 5% rather than 3%) and blast those reps. ---
**Author:** ![sje062](https://avatars.discourse-cdn.com/v4/letter/s/e8c25b/32.png) [@sje062](https://forum.mothur.org/u/sje062)\ **Post date:** [May 23, 2017, 7:32pm UTC](https://forum.mothur.org/t/classification/3263/5 "2017-05-23T19:32:19Z")
Thanks again for comment. Samples are coral. That OTU is just Bacteria unknowns, label 1. I could try for example label 3 with the phylotype command and see what happens. ---
**Author:** ![Kendra](https://yyz2.discourse-cdn.com/flex036/user_avatar/forum.mothur.org/kendra/32/112_2.png) [@Kendra](https://forum.mothur.org/u/Kendra)\ **Post date:** [May 23, 2017, 9:34pm UTC](https://forum.mothur.org/t/classification/3263/6 "2017-05-23T21:34:59Z")
label 3 will be unknown ---
**Author:** ![sje062](https://avatars.discourse-cdn.com/v4/letter/s/e8c25b/32.png) [@sje062](https://forum.mothur.org/u/sje062)\ **Post date:** [May 26, 2017, 7:46am UTC](https://forum.mothur.org/t/classification/3263/7 "2017-05-26T07:46:43Z")
I am searching for a method that group 16S sequences without using an alignment or the taxonomy. ---
**Author:** ![Kendra](https://yyz2.discourse-cdn.com/flex036/user_avatar/forum.mothur.org/kendra/32/112_2.png) [@Kendra](https://forum.mothur.org/u/Kendra)\ **Post date:** [May 26, 2017, 1:37pm UTC](https://forum.mothur.org/t/classification/3263/8 "2017-05-26T13:37:39Z")
why don’t you want to align them? ---
**Author:** ![sje062](https://avatars.discourse-cdn.com/v4/letter/s/e8c25b/32.png) [@sje062](https://forum.mothur.org/u/sje062)\ **Post date:** [May 26, 2017, 4:26pm UTC](https://forum.mothur.org/t/classification/3263/9 "2017-05-26T16:26:09Z")
They do not all overlap in the same region because from different primer sets. I like to compare sequences from different studies, downloaded from the NCBI. ---
**Author:** ![Kendra](https://yyz2.discourse-cdn.com/flex036/user_avatar/forum.mothur.org/kendra/32/112_2.png) [@Kendra](https://forum.mothur.org/u/Kendra)\ **Post date:** [May 26, 2017, 6:23pm UTC](https://forum.mothur.org/t/classification/3263/10 "2017-05-26T18:23:55Z")
ah, then you are stuck with taxonomy. You could use the approach that qiime uses where they clustered a database then match seqs to it and report the database sequence. But if your sequences aren’t in a database that approach is out.