sidle tutorial missing aligned sequence file

Hi @lam and @jwdebelius,

I am working with q2-sidle and right now some doubts has arisen for me. In the same step that @lam mentioned before:

qiime sidle reconstruct-fragment-rep-seqs \ [...]
--i-aligned-sequences database/sidle-db-aligned-sequences.qza
[...]

I could not find any place (in the "Database Preparation" or "Sequences Reconstruction" parts) inside the documentation that indicates how I can obtain the database/sidle-db-aligned-sequences.qza file. Could you, please, help me?

Thank you so much.

Regards,

Elsa

Hi @elsamdea,

The sidle-db-aligned-sequences.qza should be in the database folder you downloaded; please double check with ls. If it's not, then we can troubleshoot from there. The aligned sequence file is from the database and so it's something that gets provided during that step.

Best,
Justine

Hi @jwdebelius!

Perhaps I didn't explain myself correctly. I mean how I can build/obtain this file with my one database?

In other words, not this file: sidle-db-aligned-sequences.qza particulary, but a file with contains my database aligned-sequences.

Maybe should I re-do the alignment step (from the "Sequence Reconstruction" Section) with the reconstructed-table and the database file without separating in regions?

Sorry for the missunderstanding!

Thank you!

Elsa

Hi @elsamdea,

As always, it depends on the database. If you're working with greengenes 13_8, there should be a directory called rep_set_aligned which will have a list of files showing a percent identity; pick the one that matches your identity (probably 99). If you're using Silva 128, I think it's in the downloaded QIIME 2 folder. (Remember that tree building only works with greengenes 13_8 and Silva 128 right now. At somepoint we'll hopefully have more SEPP references, but this is where things are right now. :woman_shrugging:

Best,
Justine

2 Likes

Hi again!
I think my question is also useful in this post, because it is related to another database. So that is the reason why I am posting this here and do not in a new post.

I would love to reproduce the last step of sidle documentation's Phylogenetic Tree Reconstruction. However I could not find the rep_set_aligned folder in NCBI 16S database or neither the SEPP references file. That’s the crux of the matter.

I have found some post as:

But didn't find a solution.

Maybe I need to merge-seqs? What approximation could I use?

Any help would be appreciated!! :slight_smile:

Thank you so much!

Elsa

Hi @elsamdea,

I hope you're well! :wave:

As I think is documented in the Sidle docs (and if not I will add it): tree reconstruction only works for databases that have SEPP reference trees. This means you can reconstruct using Silva 128 or Greengenes 13_8 if you want a tree. If you'd like to build the backbone for fragment insertion with another database or database version, you're able to do that. I don't think there's a clean solution at the moment; the last thread i can find on building is from last year.

there's also a github issue:

Best,
Justine

1 Like

Hi @jwdebelius!

I am good! How have you been? :slight_smile:

I might be wrong, but I think I did not read that in the Sidle docs. But you are right, you wrote that in this post, in the answer 4.

I am so sorry!! I was confusing about SEPP references. But this explanation has resolved my doubts!!

Also, I will keep an eye on the threads you have recommended me!

Thank you so much again :blush:

Best,

Elsa

Dear Elsa and Justine,

I was confused by this last part of the tutorial. Would it be enough to align the sequences in the DB (Silva in my case) by myself in qiime? Where can I find the SEPP reference tree?

Thanks in advance,

Nadia

2 Likes

Hi @nandreani,

It sounds like I need to make it more clear in the documentation :slightly_smiling_face:.

You need the aligned sequences for hte database; I think for Silva 128, they can be downloaded from the Silva website. The insertion scaffold is available from the resources page.

Best,
Justine

1 Like

thank you very much Justine!

Nadia

1 Like