SIFTS alignment fails in cases of missing PDB residues

[ChrisMoth]

When a SIFTS "all_isoforms" restapi call aligns to missing PDB resdues (vs. transcript sequence), SIFTS does not break out segments. This is reported to Sidfts as issue [PDBe #775139] Poor alignment of 7ktr chain A with uniprot seq Q9Y4A5-2

Most likely, they will soon at a uniprot_segments API call that includes support for non-canonical uniprot IDs.

Excerpt from the email exchange with them:

SOME "all_isoforms" sifts mappings have multiple pairs for non-canonical isoforms. Consider mappings/all_isoforms/W5XKT8-2 and mappings/all_isoforms/W5XKT8-3

https://www.ebi.ac.uk/pdbe/api/mappings/all_isoforms/W5XKT8-2
https://www.ebi.ac.uk/pdbe/api/mappings/all_isoforms/W5XKT8-3

For these the returned JSON returns TWO segments:
7ta2.PDB chain A residues 48-52 -> uniprot 73-77 (5 residue segment mapped)
7ta2.PDB chain A residues 1-46 -> uniprot 27-72 (46 residue segment mapped)

The SIFTS API did NOT return a single mapping 1-52->27-77 to match the behavior of Q9Y4A5-2 and 7ktr.pdb
....

[ChrisMoth]

One can demonstrate 7000 examples of the api returning segments
select * from (select uniprot_acc, pdbid, mapping_pdb_chain, count(uniprot_acc) as cnt from pdbmap_v14.sifts_mappings_pdb_uniprot_all_isoforms group by uniprot_acc, pdbid, mapping_pdb_chain) as t where t.cnt > 1;