Variant depth issues

[cmaceves]

addressing issues #85 and issues #126
ivar variants currently calculates frequency using ungapped depth, however consensus is called using gapped depths.
for consistency, we now use gapped depths when calling variants.

[PovilasMat]

Hi @cmaceves, I havent tested the change. However, in our experience variant calling for covid was correctly identifying deletions while consensus calling was incorrect and calling leading or trailing Ns within deletions (more often than not just a single nucleotide which caused frameshifts). So while changing variant calling to act like consensus calling makes it consistent, for many of us that doesn't provide a solution.

[cmaceves]

hey! could you please provide some example data with the exact commands @PovilasMat ? because in the sample data I had/was using, those trailing N's resulted from an actually ambiguity. for example at the trailing positions in my example data, the deletion occurred at ~50% and NT's matching the reference occurred at ~50%, leading to N's being called

[PovilasMat]

@cmaceves , I need to figure out things with legal department to be allowed to share samples. However, in the mean time, I would be willing to meet and show some examples.

[cmaceves]

@PovilasMat sure, you can reach me at [email protected] to schedule something!

[joshuailevy]

bumping this :)