sepp should reject distant sequences

[sjanssen2]

Hi @smirarab,
can you please have a look at an issue in our qiime2 forum. A user found a situation in which SEPP inserts fragments even though they are only very remotely similar to anything in the reference.
Thanks,
Stefan

[smirarab]

Yes, it currently is not trying to filter out wrong sequences. We can try to add an option (not default) were poorly aligned sequences are not placed. Would that help?

…

On Tue, Jul 9, 2019 at 11:04 AM Stefan Janssen ***@***.***> wrote: Hi @smirarab <https://github.com/smirarab>, can you please have a look at an issue in our qiime2 forum <https://forum.qiime2.org/t/non-16s-seqs-inserted-into-gg13-8-ref-tree-by-sepp/10561>. A user found a situation in which SEPP inserts fragments even though they are only very remotely similar to anything in the reference. Thanks, Stefan — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#73?email_source=notifications&email_token=AAGJXODRGX45AQXDACZTEDTP6THMTA5CNFSM4H7IAFA2YY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4G6FJO5Q>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAGJXOC2G56NCCBTA4PZG4DP6THMTANCNFSM4H7IAFAQ> .

-- Siavash Mirarab

[sjanssen2]

Our initial use case for SEPP was in combination with Deblur, which automatically performs this filtering of distant sequences. One can alternatively use DADA2 or other sOTU methods which might not do this filtering.
However, I think it would make more sense to have this filtering capability as its own module rather than adding it to SEPP.

[smirarab]

I think I agree.

…

On Thu, Jul 11, 2019 at 11:51 PM Stefan Janssen ***@***.***> wrote: Our initial use case for SEPP was in combination with Deblur, which automatically performs this filtering of distant sequences. One can alternatively use DADA2 or other sOTU methods which might not do this filtering. However, I think it would make more sense to have this filtering capability as its own module rather than adding it to SEPP. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#73?email_source=notifications&email_token=AAGJXOF4XUXNJQ6DNDTI4JTP7ASYBA5CNFSM4H7IAFA2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODZY3UKI#issuecomment-510769705>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAGJXOHIEEVPK23C5DXOHZLP7ASYBANCNFSM4H7IAFAQ> .

-- Siavash Mirarab