Incorrect single-nucleotide duplication projection

[leicray]

Describe the bug A user reports that the single-nucleotide sequence variant NM_000892.3:c.451dup is predicted to be a two-nucleotide duplication (NC_000004.12:g.186236902_186236903dup) when projected onto GRCh38. The warning NM_000892.3:c.451dupT cannot be mapped directly to genome build GRCh38 is also generated.

The sequence variant with respect to versions 4 and 5 of the reference sequence validates correctly.

This might be due to an error in the alignment of the transcript NM_000892.3 to GRCh38.

[Peter-J-Freeman]

The key looks to me the following warning when I try to validate NM_000892.3:c.451dupT sirectly to GRCh38

            ```NM_000892.3:c.451dupT cannot be mapped directly to genome build GRCh38```

I need to look at how the liftover is being performed in this particular instance. It may well not be ab alignment issue, but a workflow issue which for this particular variant is problematic.

[John-F-Wagstaff]

OK, so there are actually 2 issues here.

The first lower priority issue is that this variant is lifted over using the intergenic lift-over code as there is no mapping for NM_000892.3 on GRCh38, despite the fact that other transcripts are available which map to both GRCh38 and GRCh37. The complexity of fixing this problem is relatively high, and any fix for this issue may cause unexpected behaviour in some cases so I have left as is for now. @Peter-J-Freeman do we want an extra feature request type issue opened for this?

The second issue is that intergenic lift over of dup type variants does not quite work right. I have built and tested a fix for this now which should solve this issue. @Peter-J-Freeman assuming I have not messed up with this this should be available in develop now, can you have a quick check of my logic on this, and pull into main if/when appropriate?

[Peter-J-Freeman]

@John-F-Wagstaff . Which branch is your fix on?

[John-F-Wagstaff]

https://github.com/openvar/variantValidator/tree/develop I had intended to push it to a separate branch, and then send a pull request, but I forgot to switch branches before submitting the patches and only realised after I pushed :-(

[Peter-J-Freeman]

no worries, I'll pull into my dev branch

[Peter-J-Freeman]

All looks good. closing