Fragment insertion tree - tips dont correspont to OTU IDs

Hi All,

I’ve been going down this rabbit hole and hoping you can dig me out - I’ve used q2-fragment-insertion to construct a tree and exported that tree (along with my feature table) and am trying to use Phylofacttorization (Note - these are data that were initially downloaded from Qiita and imported to Qiime2).

for some reason, looking at tree tip.labels (after exporting) gives me a number and not the OTU ID (which is the sequence rather than the hash). Any idea what I’m doing wrong?


Thanks for posting, @Gil_Sharon!

@Stefan any ideas on this?

Hi @Gil_Sharon,

Let me start saying that deblur outputs can have the sequence hashes or the sequences themselves as ids; the difference is a parameter. Qiita uses the deblur sequence as ids (for the deblur results) as the feature of hash sequences ids was added after deblur was added to Qiita. Now, fragment insertion will insert the new fragments into a tree, in this case 99% GG 13_8. Thus, the final tree will have the original GG OTU ids (numbers) and the new inserted sequences.

Hope this helps.


@antgonza Thanks! this makes perfect sense now.
this change might make exporting a bit of an issue - following SEPP, there is a discrepancy between the tree tips and feature IDs in the OTU table (I suppose Qiime2 solves it in the back-end). Can one output a new OTUtable and taxonomy (and repset, although this can be pulled out of GG) files with updated feature IDs? couldnt find an option in q2-Fragment Insertion.


@Gil_Sharon, I don’t follow which one is the possible discrepancy but I think your concern is that now your feature table will have a subset of the features that are in your tree. If that’s the case, that’s not a problem at least for the regular stuff that we do with the tree (like UniFrac or Faith’s PD) as those not included tips in the feature table are ignored, which should be the case for all methods I can think of. Again, I guess it depends on your algorithm.

Just to be clear, the other way around, might be a problem (features in your feature table that are not in the tree), which we normally consider noisy sequences but we also know that nothing is perfect. The current thing we do is to simply remove them but this is a case by case decision.

Perhaps useful, here is a list of commands for Transferring Qiita Artifacts to Qiime2.

Thanks, @antgonza. I exported my OTU table and SEPP tree (and taxonomy) so I can load them into R for phylofactor. The tree$tip.labels should be exactly the same as the OTU IDs - but while the tree labels are GG OTU #s, the OTU table’s are the deblur sequences.

Thanks for link to the tutorial! I downloaded the data before SEPP was running in Qiita, so ran it offline on the data I downloaded before. The code should in the tutorial would probably fix this issue.


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