Separate multiple error messages for the LOVD endpoint.

[ifokkema]

Is your feature request related to a problem? Please describe.
The "normal" VV endpoint uses an array to return multiple issues. For instance, NM_000088.3:c.589delG returns:

{
    "validation_warnings": [
        "Removing redundant reference bases from variant description",
        "NM_000088.3:c.589del automapped to NM_000088.3:c.590del",
        "TranscriptVersionWarning: A more recent version of the selected reference sequence NM_000088.3 is available for genome build GRCh37 (NM_000088.4)"
    ]
}

That is very useful. However, the LOVD endpoint does not do this. For instance, for NC_000017.10:g.100_110del10, a string is returned:

{
    "genomic_variant_error": "Length implied by coordinates must equal sequence deletion length, Trailing digits are not permitted in HGVS variant descriptions, Refer to http://varnomen.hgvs.org/recommendations/DNA/variant/"
}

This causes some issues because we parse the string and ignore some things. For instance, we ignore "Refer to http://...", but that means we'd ignore the entire string now.

Describe the solution you'd like
Since we can't turn that into an array without breaking existing implementations of the API, for now, we're stuck with using a string. But perhaps we can use another trick. Do VV errors or warnings ever use semicolons in the text? If not, can we separate the individual messages with semicolons? Or can you think of some other way that we can separate the individual messages?

Describe alternatives you've considered
I assume that I can't split on commas, because I assume they are used in VV errors. I don't have data that confirms this, though. I'm not sure if you have some kind of list of possible errors and warnings?

Additional context
I realized here that VV links to the old HGVS website in "Refer to http://varnomen.hgvs.org/recommendations/DNA/variant/". You may want to fix that, too 😉

[Peter-J-Freeman]

@ifokkema, the dificulty as always is the HGVS use of every character on earth, then some. The ; is likely used in the nomenclature, so it is not an available option. How about an &???

[Peter-J-Freeman]

p.s. will update the web links in the next version. Its not necessarily a small job. We link to different pages so need to check all the links manually to replace them.

[ifokkema]

@ifokkema, the dificulty as always is the HGVS use of every character on earth, then some. The ; is likely used in the nomenclature, so it is not an available option. How about an &???

Right, the given variant description is sometimes also in the message... It could also be the combination of a semicolon and a space, like: "Message1; Message2", or even "Message1.; Message2.", but in that latter case, all errors and warnings should end with a period, and now they never do, as far as I know. Anyway, splitting on "; " is still an option, I think, or we go with "&" like you proposed, maybe even two of them surrounded by spaces: "Message1 && Message 2".

p.s. will update the web links in the next version. Its not necessarily a small job. We link to different pages so need to check all the links manually to replace them.

Ah, I see... some may be easy, but not all, indeed. In fact, the "index" pages for DNA, RNA, and protein levels have no new counterparts; they have no replacement links.