Add sequence source

translate2modelcif.py

@@ -7,6 +7,7 @@ import sys
 
 import requests
 
+import ihm
 import modelcif
 import modelcif.dumper
 import modelcif.reference
@@ -236,12 +237,18 @@ def _store_as_modelcif(interaction_name, data_json, file_prfx):
         id=interaction_name.upper(),
         model_details=data_json["model_details"],
     )
-    # create target entities, references, ...
+    # create target entities, references, source...
+    # for source we assume all chains come from the same taxon
+    source = ihm.source.Natural(
+        ncbi_taxonomy_id=data_json["target_entities"][0]["up_ncbi_taxid"],
+        scientific_name=data_json["target_entities"][0]["up_organism"],
+    )
     for cif_ent in data_json["target_entities"]:
         # ToDo [input]: Get entity description
         mdlcif_ent = modelcif.Entity(
             cif_ent["cif_sequence"],
             description="entity.pdbx_description",
+            source=source,
             references=[
                 modelcif.reference.UniProt(
                     cif_ent["up_id"],
@@ -258,8 +265,10 @@ def _store_as_modelcif(interaction_name, data_json, file_prfx):
                 )
             ],
         )
+
         system.entities.append(mdlcif_ent)
         system.target_entities.append(mdlcif_ent)
+
     # audit_authors
     system.authors.extend(data_json["audit_authors"])
 
@@ -327,5 +336,5 @@ if __name__ == "__main__":
 #  LocalWords:  struct cif utf mmcif fh datetime ost io ToDo chn lst rn idx aa
 #  LocalWords:  sqe olc ACDEFGHIKLMNPQRSTVWY RuntimeError upkb txt pylint iter
 #  LocalWords:  rspns unicode startswith sline len elif NCBI TaxID ncbi taxid
-#  LocalWords:  seqlen crc ISOFORM DT dt flds isoforms isoform ent LoadPDB
+#  LocalWords:  seqlen crc ISOFORM DT dt flds isoforms isoform ent LoadPDB ihm
 #  LocalWords:  mdlcf mdlcif