Liftover fails for GRCh37 variants where the strand differs from GRCh38

[jarbesfeld]

Describe the bug

There are genes and transcripts, such as NBPF20, where the strand that the gene occurs on changes between reference builds. Currently, Cool-Seq-Tools fails to get the lifted over MANE transcript for variants that occur on these genes that are described on GRCh37.

Steps to reproduce

await f.cool_seq_tool.mane_transcript.get_mane_transcript(
    "NC_000001.10",
    148024875,
    148024875,
    "g",
)

Expected behavior

GenomicRepresentation(pos=(145402330, 145402331), mane_genes=[ManeGeneData(ncbi_gene_id=100288142, hgnc_id=32000, symbol='NBPF20', status=[<ManeStatus.SELECT: 'mane_select'>])], status='grch38', ac='NC_000001.10')

Current behavior

None is returned

Possible reason(s)

No response

Suggested fix

No response

Branch, commit, and/or version

branch: main

Screenshots

No response

Environment details

Mac M4 Pro

Additional details

No response

Contribution

None

[jarbesfeld]

@ahwagner Following up on this, as this was the cause of why we believe some validly-formatted SNVs failed normalization. An example SNV that failed normalization was NC_000001.10:g.148024875C>T. In the ClinGen Allele Registry, this lifts over the following variant: NC_000001.11:g.145402331C>T.

However, when supplying position 48024875 on chr 1 into the broad liftover tool, we see a different mapped position on GRCh38:

When trying to lift this position back to GRCh37, an error is returned:

Is there a policy that you would recommend following for these cases?