genomic / cdna coord conversion of upstream/downstream regions on NR transcripts

[rolfschr]

Hi,

I found some issues with the genomic / cdna coord conversion for NR transcripts. I base my findings on comparing the results from pyhgvs with Alamut Visual (which in turn is consistent with https://mutalyzer.nl/position-converter). Specifically, I saw that upstream (befor first exon) and downstream (after last exon) cdna coords were copmletely off for non-coding (NR) transcripts. Exonic and intronic regions are fine. I believe the issue stems from trying to use the (non-existing) start (or end) position of the cds regions as an offset. I have added some unittests which fail.

I have also supplied a patch which makes the unittests pass. However, I believe this is not the way one would implement genomic_to_cdna_coord(). It's more like a hint. I also added a crappy hack for a very specific scenario in cdna_to_genomic_coord() where I did not find how to do it "more" correctly.

As an example:

chr13:g.70681018del on ATXN8OS NR_002717.2 (upstream) is coded as NR_002717.2:n.-327 by Alamut/Mutalyzer but pygvs.genomic_to_cdna_coord('NR_002717.2', 70681018) returns -326. chr13:g.70714013del (same tx, downstream) should be *128 but is 1600.

[davmlaw]

I think "NR_002717.2:n.-327" violates the HGVS spec:

https://varnomen.hgvs.org/bg-material/numbering/

non-coding DNA reference sequences

It is not allowed to describe variants in nucleotides which are not covered by the transcript using a non-coding DNA reference sequence