OTU sequence

[ZarulHanifah]

Hey NINJA-OPS developers,

Thank you for this wonderful software. It is very quick and simple.

May I ask, does NINJA actually produce an OTU representative fasta file? Because I need to construct a phylogenetic tree based on the OTU sequences (for phylogenetic calculation stuff in QIIME)

[GabeAl]

Hi! Thanks for using NINJA-OPS! Because NINJA-OPS uses a standard closed-reference database by default, in this case Greengenes 13_8, the Greengenes reference tree should be used for phylogenetic analysis. Since you mentioned you were using Qiime, you probably have this tree already--let me know if you don't and I can attach it here. Thanks, Gabe On Sep 5, 2017 6:38 AM, "ZarulSaurus" <notifications@github.com> wrote: Hey NINJA-OPS developers, Thank you for this wonderful software. It is very quick and simple. May I ask, does NINJA actually produce an OTU representative fasta file? Because I need to construct a phylogenetic tree based on the OTU sequences (for phylogenetic calculation stuff in QIIME) — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#17>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AHrXBhbd_wFQsAxi2zXURe8XopT2bWpYks5sfTKagaJpZM4PM0YJ> .

[ZarulHanifah]

Thank you very much for the prompt reply!

Now I see; so I am familiar with QIIME, but only for the downstream analysis part (For my previous work, I used UParse for OTU clustering). I only noticed just now that the OTU ID was named from the aligned GreenGenes reference sequences as well.

From print_qiime_config.py, all the default files have been listed out. But I didnt see any reference tree. Should I extract the reference OTUs found in my samples and make a tree, or is there a reference tree that I should be using? If there is a GreenGene reference tree, can you share it here? That would be awesome.

Also, I side-question, if I am planning to do open-reference OTU picking, I assume that I just have to run NINJA, then extract the sequences from ninja_fail.txt and do de-novo OTU picking from there?

Thank you so much :)

[GabeAl]

Sure, here's the tree! 👍

97_otus.zip

Yes, that's one reasonable way to do closed ref. If there are too many failures, I'd lower the id to 95% or so (the -s or --similarity option in NINJA-OPS) and that'll hopefully pare down the failures to a small enough number for reasonable de novo on them. Be sure to merge the OTU tables using QIIME's command (one mistake I've seen is to "cat" the files together directly).

A more complicated way would be to start the open_ref pipeline with some other aligner, but replace the first stage with NINJA-OPS. The open-ref pipeline in QIIME does some funky things with resampling and re-aligning to the random subsample, so it's a little different than pure de novo. I think if you have the computational resources, pure de novo on the failures might be as good or better in some cases... my conjecture.

Cheerio!