Frameshift detection

[hoelzer]

@replikation following up the discussion on BC: I think it's reasonable to put a terminal print warning about Omicron sequences missing many Spike mutations and showing sup basecalling and/or switching to Nanopolish as options to maybe fix that.

However, instead, we could also implement a general check for frameshifts (FS). For example, this could be done via https://gitlab.com/s.fuchs/covsonar in two easy steps:

All reconstructed consensus sequences can be added to a covSonar database:

sonar.py add -f genomes.fasta --db mydb --cpus 8

Then, we can query this database via

sonar.py match --db mydb --only_frameshifts | awk 'BEGIN{FS=","};{print $1}' | grep -v accession > ids-frameshift.txt

which will give back all sequence IDs that have a frameshift.

Now, we could additionally mark them in the report and/or print a message that one should be aware of that and maybe try basecalling with a higher accuracy model or switching to Nanopolish/Medaka. Or at least investigate the sequences if the frameshift is actually true.

This would also help people w/ subsequent analyses and e,g, GISAID upload, ...

What do you think?

[MarieLataretu]

(nextclade also reports the frameshifts in the frameShifts column)

[hoelzer]

(nextclade also reports the frameshifts in the frameShifts color)

ah true - that we already have. ;) would be even easier than introducing another tool.

[MarieLataretu]

For the report I see two options:

• report frame shifts in a additional column and highlight it like mutations on the Spike
• don't report them, but highlight the sample (see Report classification as VOI, VOC, ... of lineages in results table #190)

[MarieLataretu]

I think, an additional column is the way to go.
@RaverJay can you add the frameShifts column to the report, if you all agree @replikation @hoelzer ?

It's on amino acid level, but good enough for now.