Report Insertions as well about porecov

[hoelzer]

I think it's difficult to include in the final report at least. There can be various insertions also based on seq errors, used protocols, ...

The question would be also how the newest Nextstrain output looks like in this context?

My last idea was to at least add a column for the totalInsertions so that one can easily spot weirdos for further checks

[corneliusroemer]

In the upcoming release (today or tomorrow) Nextclade/align will output aa insertions. Maybe this is still relevant?

[RaverJay]

Well we are reporting amino acid changes, so these columns:
"aaSubstitutions"
"aaDeletions"

but there is no "aaInsertions" at this time.

Also the "insertions" column is empty in all the runs I just checked

[hoelzer]

Yes, I think insertions are rare. Let me check if I can find an example Nextclade output w/ insertions.

[hoelzer]

Okay, so it seems insertions are only reported on nucleotide level. E.g. I just checked some sequences and selected some different outputs:

totalInsertions insertions
17      21534:A,27373:CTTTCGATCTCT,27380:C,27384:CTC
16      27373:CTTTCGATCTCT,27380:C,27384:CTC
10      9627-9631,20528-20531,22492,28967-28969
28      1569:C,1574:AGAGCTAG,27373:CTTTCGATCTCT,27380:C,27384:CTC,28250:CTG
many    matches"""
248     7010-7049,9543-9561,14331-14515,29760-29767
1       28250:G
17      39,19286,29602,29605,29631,29706,29752,29772,29779,29785,29792-29794,29797,29806,29866,29868-29870

It seems the output format can vary a lot. So I am not sure if it is so meaningful to add this. Maybe just information if there are insertions or not (totalInsertions)?

[replikation]

• @hoelzer @RaverJay status?

[hoelzer]

@replikation @RaverJay actually this is getting now important with Omicron, especially because of an larger insertion in the Spike:

S:R214REPE

Currently, this is just not shown in the report which can lead to misleading results.

Not sure what's best to solve this, I think currently we could still only extract such information from Nextclade.

Another way would be to use our tool covsonar to basically call all substitutions, insertions and deletions with one tool and include this into the report... https://gitlab.com/s.fuchs/covsonar

To do so, we would need a process that generates the covsonar database for all genomes and then we could easily extract all information.

[RaverJay]

Yeah this is bad

Problem is, Nextclade output is still missing a 'aaInsertions' column
(there is only: substitutions deletions insertions aaSubstitutions aaDeletions)

We could of course calculate it from 'insertions' though

@replikation what do you think, add or switch to covsonar?

[hoelzer]

Ahh, we would then need a converter from nt insertions reported by Nextclade to aa insertions. Something like

https://codon2nucleotide.theo.io/

CovSonar might solve that but here the only weak point is that the tool is also under development still (currently people are on a CovSonar2 version) so there might be certain changes. And it's not tested so extensively like Nextclade, etc...

[RaverJay]

Nextclade might take too long to add this: nextstrain/nextclade#319

Maybe we should implement it ourselves at this time - just 'borrow' the code from codon2nucleotide and translate to AAs

[hoelzer]

Yes, seems fair to me. Then we still rely on Nextclade that anyway runs and just translate the nt insertions for the report. I thin having the code from https://github.com/theosanderson/codon2nucleotide as a small script in bin should do the trick?

[RaverJay]

It's a little bit more complicated than that, see PR

[replikation]

@corneliusroemer thanks for informing us