Do not seed cyclopiazonic-acid CZA PDB rows on CHEBI:22450

[realmarcin]

Blocked rows

The PubChem PDB-derived rows for the NaturalProductMech CHEBI:22450 record point at SERCA structures that are protein-groundable but ligand-mismatched:

• data/raw/pubchem_bioassay.tsv:7982, AID 1811, target 2O9J_A, PMID:17259168
• data/raw/pubchem_bioassay.tsv:7983, AID 1811, target 2OA0_A, PMID:17259168
• data/raw/pubchem_bioassay.tsv:7984, AID 977610, Ki = 0.12 uM, target 2O9J_A, PMID:17259168
• data/raw/pubchem_bioassay.tsv:7985, AID 977610, Ki = 0.12 uM, target 2OA0_A, PMID:17259168

Why this should not become a graph yet

The record is data/natural_products/unclassified/6ar-11as-11br-10-acetyl-11-hydroxy-7-7-dimethyl-2-6-6a-7-11a-11b-hexahydro-9h-py.yaml:

• identifier: CHEBI:22450
• record Standard InChIKey: SZINUGQCTHLQAZ-DQYPLSBCSA-N
• source PubChem CID: 54682463

The PDB evidence itself is valid for a rabbit SERCA protein:

• RCSB 2O9J primary DOI: 10.1074/jbc.M611653200
• RCSB 2O9J PDB DOI: 10.2210/pdb2o9j/pdb
• RCSB 2OA0 PDB DOI: 10.2210/pdb2oa0/pdb
• PDBe maps both 2O9J_A and 2OA0_A to reviewed UniProtKB:P04191 / AT2A1_RABIT

However, both PDB entries contain CZA, and the RCSB CZA chem comp has InChIKey=RLOAZVAJNNPPDI-DQYPLSBCSA-N and related PubChem CID 65261, not SZINUGQCTHLQAZ-DQYPLSBCSA-N.

Suggested resolution

Do one of these before adding a cyclopiazonic-acid-serca-binding graph:

• add a separate exact record for PubChem 65261 / PDB CZA and attach the 2O9J/2OA0 rows there, or
• model the tautomer/protomer relationship explicitly enough that a graph on CHEBI:22450 is not an exact-structure overstatement.

Acceptance checks:

• exact ignored-file-inclusive searches for P04191, 10.1074/jbc.M611653200, 10.2210/pdb2o9j/pdb, and 10.2210/pdb2oa0/pdb remain clean except for the intentional graph or blocker
• the PDB ligand Standard InChIKey matches the record that receives the SERCA causal graph