Decide how to register multi-variant cis-phased blocks with ClinGen

[bencap]

Context

get_hgvs_from_post_mapped (src/mavedb/lib/variants.py) now supports an opt-in combine_cis=True that collapses a multi-variant cis-phased block (VRS Haplotype / CisPhasedBlock) into a single bracketed HGVS expression, e.g. NC_000001.11:g.[123A>G;125T>C], via the new join_cis_phased_hgvs helper in src/mavedb/lib/hgvs.py.

We have enabled combine_cis=True in:

• the variant mapping job (hgvs_assay_level + Allele.hgvs_* population)
• the score-set response/export view models (post_mapped_hgvs_g / post_mapped_hgvs_p fallbacks)

We have deliberately left it disabled for ClinGen Allele Registry submission (src/mavedb/worker/jobs/external_services/clingen.py, both call sites). Today these variants return no HGVS and are recorded as MISSING_IDENTIFIER failures and skipped.

The question

A multi-variant cis-phased block has no single Canonical Allele ID, and it is unclear whether the ClinGen Allele Registry accepts a bracketed cis-phased HGVS expression (g.[a;b]) at all, or whether these should instead be:

• registered per-component (one CAID per member), then associated, or
• represented some other way the registry prefers, or
• intentionally left unregistered.

Action

Work with ClinGen to confirm the intended handling, then either enable combine_cis=True at the ClinGen call sites (and adjust submission/annotation accordingly) or formalize the skip. Update the inline notes referencing this issue once resolved.