Notice partial gene deletion that spans start codon about drprg HOT 8 CLOSED

In particular, this issue is concerned with solving problem 2 above.

We can detect whole gene deletions, but not partial deletions which knock out the start codon - which effectively amount to the same thing.

I've seen this issue in both pncA and katG.

from drprg.

The implementation of this feature will likely need to change when/if iqbal-lab-org/pandora#316 is closed.

from drprg.

Do you mean that there is a potential fix even without a fix for iqbal-lab-org/pandora#316 ?

from drprg.

Well noticing failed variants that span the start codon would be a kind of band-aid fix. The proper fix would be the resolution of that pandora issue

from drprg.

Should we also be detecting when the stop is lost? There are two INH FNs that we miss because we don't detect stop loss and tbprofiler calls stop loss. We have null genotypes spanning the stop codon in both of these samples.

from drprg.

My guess is yes we should

from drprg.

I made a change to the partial gene deletion code and also removed the GT CONF filter. The (Illumina) diff I get from these changes is

Most of these FPs are also FPs on tbprofiler also.

Regarding the stop lost stuff, Miranda made a good point, maybe we just flag it as an unknown mutation?

from drprg.

Good idea, flag as unknown seems safest.

from drprg.