Make vrs-python usable for genomic variant normalization without installing postgresql dependencies
If you want to normalize any variants using vrs-python right now you need to install psycopg2 and a C compiler and host psql libraries. Genomic sequence variants do not require UTA though and there should be an optional dependency target to install only what is needed for those kinds of variants, excluding all postgresql dependencies.
Related https://github.com/broadinstitute/seqr-loading-pipelines/issues/1089
Also a good reminder that the gnomad python library is still pinned to 0.8.4
https://github.com/broadinstitute/gnomad_methods/blob/v0.8.2/requirements.txt#L2
This may be difficult since hgvs requires a postgres library and we need hgvs parsing even for genomic variants. It is possible but unlikely for the hgvs package to make postgres optional. It would require significant code changes.