Improve Rhea import (#168)

src/pyobo/sources/famplex.py

@@ -151,9 +151,11 @@ def _get_xref_df(version: str) -> Mapping[str, List[Reference]]:
     }
     xrefs_df[0] = xrefs_df[0].map(lambda s: ns_remapping.get(s, s))
     xrefs_df[1] = [
-        bioregistry.standardize_identifier(xref_prefix, xref_identifier)
-        if xref_prefix != "nextprot.family"
-        else xref_identifier[len("FA:") :]
+        (
+            bioregistry.standardize_identifier(xref_prefix, xref_identifier)
+            if xref_prefix != "nextprot.family"
+            else xref_identifier[len("FA:") :]
+        )
         for xref_prefix, xref_identifier in xrefs_df[[0, 1]].values
     ]
 

src/pyobo/sources/rhea.py

@@ -3,31 +3,51 @@
 """Converter for Rhea."""
 
 import logging
-from typing import Iterable
+from typing import TYPE_CHECKING, Dict, Iterable, Optional
 
+import bioversions
 import pystow
 
 from pyobo.struct import Obo, Reference, Term
 from pyobo.struct.typedef import (
+    TypeDef,
+    enabled_by,
     has_bidirectional_reaction,
+    has_input,
     has_left_to_right_reaction,
+    has_output,
+    has_participant,
     has_right_to_left_reaction,
+    reaction_enabled_by_molecular_function,
 )
 from pyobo.utils.path import ensure_df
 
+if TYPE_CHECKING:
+    import rdflib
+
 __all__ = [
     "RheaGetter",
 ]
 
 logger = logging.getLogger(__name__)
 PREFIX = "rhea"
+RHEA_RDF_GZ_URL = "ftp://ftp.expasy.org/databases/rhea/rdf/rhea.rdf.gz"
 
 
 class RheaGetter(Obo):
     """An ontology representation of Rhea's chemical reaction database."""
 
     ontology = bioversions_key = PREFIX
-    typedefs = [has_left_to_right_reaction, has_bidirectional_reaction, has_right_to_left_reaction]
+    typedefs = [
+        has_left_to_right_reaction,
+        has_bidirectional_reaction,
+        has_right_to_left_reaction,
+        enabled_by,
+        has_input,
+        has_output,
+        has_participant,
+        reaction_enabled_by_molecular_function,
+    ]
 
     def iter_terms(self, force: bool = False) -> Iterable[Term]:
         """Iterate over terms in the ontology."""
@@ -39,25 +59,54 @@ def get_obo(force: bool = False) -> Obo:
     return RheaGetter(force=force)
 
 
+def ensure_rhea_rdf(version: Optional[str] = None, force: bool = False) -> "rdflib.Graph":
+    """Get the Rhea RDF graph."""
+    # see docs: https://ftp.expasy.org/databases/rhea/rdf/rhea_rdf_documentation.pdf
+    if version is None:
+        version = bioversions.get_version(PREFIX)
+    return pystow.ensure_rdf(
+        "pyobo",
+        "raw",
+        PREFIX,
+        version,
+        url=RHEA_RDF_GZ_URL,
+        force=force,
+        parse_kwargs=dict(format="xml"),
+    )
+
+
+def _get_lr_name(name: str) -> str:
+    return name.replace(" = ", " => ")
+
+
+def _get_rl_name(name: str) -> str:
+    left, right = name.split(" = ", 1)
+    return f"{right} => {left}"
+
+
+def _get_bi_name(name: str) -> str:
+    return name.replace(" = ", " <=> ")
+
+
 def iter_terms(version: str, force: bool = False) -> Iterable[Term]:
     """Iterate over terms in Rhea."""
-    url = "ftp://ftp.expasy.org/databases/rhea/rdf/rhea.rdf.gz"
-    graph = pystow.ensure_rdf(
-        "pyobo", "raw", PREFIX, version, url=url, force=force, parse_kwargs=dict(format="xml")
-    )
+    graph = ensure_rhea_rdf(version=version, force=force)
     result = graph.query(
-        """
-    PREFIX rh:<http://rdf.rhea-db.org/>
-    SELECT ?reaction ?reactionId ?reactionLabel WHERE {
-      ?reaction rdfs:subClassOf rh:Reaction .
-      ?reaction rh:id ?reactionId .
-      ?reaction rdfs:label ?reactionLabel .
-    }
+        """\
+        PREFIX rh:<http://rdf.rhea-db.org/>
+        SELECT ?reaction ?reactionId ?reactionLabel WHERE {
+          ?reaction rdfs:subClassOf rh:Reaction ;
+                    rh:id ?reactionId ;
+                    rdfs:label ?reactionLabel .
+        }
     """
     )
-    names = {str(identifier): name for _, identifier, name in result}
+    names = {str(identifier): str(name) for _, identifier, name in result}
 
-    terms = {}
+    terms: Dict[str, Term] = {}
+    master_to_left: Dict[str, str] = {}
+    master_to_right: Dict[str, str] = {}
+    master_to_bi: Dict[str, str] = {}
 
     directions = ensure_df(
         PREFIX,
@@ -66,12 +115,16 @@ def iter_terms(version: str, force: bool = False) -> Iterable[Term]:
         force=force,
     )
     for master, lr, rl, bi in directions.values:
-        terms[master] = Term(
-            reference=Reference(prefix=PREFIX, identifier=master, name=names.get(master))
-        )
-        terms[lr] = Term(reference=Reference(prefix=PREFIX, identifier=lr, name=names.get(lr)))
-        terms[rl] = Term(reference=Reference(prefix=PREFIX, identifier=rl, name=names.get(rl)))
-        terms[bi] = Term(reference=Reference(prefix=PREFIX, identifier=bi, name=names.get(bi)))
+        master_to_left[master] = lr
+        master_to_right[master] = rl
+        master_to_bi[master] = bi
+
+        name = names[master]
+
+        terms[master] = Term(reference=Reference(prefix=PREFIX, identifier=master, name=name))
+        terms[lr] = Term(reference=Reference(prefix=PREFIX, identifier=lr, name=_get_lr_name(name)))
+        terms[rl] = Term(reference=Reference(prefix=PREFIX, identifier=rl, name=_get_rl_name(name)))
+        terms[bi] = Term(reference=Reference(prefix=PREFIX, identifier=bi, name=_get_bi_name(name)))
 
         terms[master].append_relationship(has_left_to_right_reaction, terms[lr])
         terms[master].append_relationship(has_right_to_left_reaction, terms[rl])
@@ -80,6 +133,38 @@ def iter_terms(version: str, force: bool = False) -> Iterable[Term]:
         terms[rl].append_parent(terms[master])
         terms[bi].append_parent(terms[master])
 
+    # inspired by https://github.com/geneontology/go-ontology/blob/master/src/sparql/construct-rhea-reactions.sparql
+    sparql = """\
+    PREFIX rh:<http://rdf.rhea-db.org/>
+    SELECT ?reactionId ?side ?chebi WHERE {
+      ?reaction rdfs:subClassOf rh:Reaction ;
+                rh:id ?reactionId .
+
+      ?reaction rh:side ?side .
+      ?side rh:contains ?participant .
+      ?participant rh:compound ?compound .
+      ?compound rh:chebi|rh:underlyingChebi|(rh:reactivePart/rh:chebi) ?chebi .
+    }
+    """
+    for master_rhea_id, side_uri, chebi_uri in graph.query(sparql):
+        master_rhea_id = str(master_rhea_id)
+        chebi_reference = Reference(
+            prefix="chebi", identifier=chebi_uri[len("http://purl.obolibrary.org/obo/CHEBI_") :]
+        )
+        side = side_uri.split("_")[-1]  # L or R
+        if side == "L":
+            left_rhea_id = master_to_left[master_rhea_id]
+            right_rhea_id = master_to_right[master_rhea_id]
+        elif side == "R":
+            left_rhea_id = master_to_right[master_rhea_id]
+            right_rhea_id = master_to_left[master_rhea_id]
+        else:
+            raise ValueError(f"Invalid side: {side_uri}")
+        terms[master_rhea_id].append_relationship(has_participant, chebi_reference)
+        terms[master_to_bi[master_rhea_id]].append_relationship(has_participant, chebi_reference)
+        terms[left_rhea_id].append_relationship(has_input, chebi_reference)
+        terms[right_rhea_id].append_relationship(has_output, chebi_reference)
+
     hierarchy = ensure_df(
         PREFIX,
         url="ftp://ftp.expasy.org/databases/rhea/tsv/rhea-relationships.tsv",
@@ -91,39 +176,59 @@ def iter_terms(version: str, force: bool = False) -> Iterable[Term]:
             raise ValueError(f"RHEA unrecognized relation: {relation}")
         terms[source].append_parent(terms[target])
 
-    for xref_prefix, url in [
-        ("ecocyc", "rhea2ecocyc"),
-        ("kegg.reaction", "rhea2kegg_reaction"),
-        ("reactome", "rhea2reactome"),
-        ("macie", "rhea2macie"),
-        ("metacyc", "rhea2metacyc"),
+    for xref_prefix, url, relation in [
+        ("ecocyc", "rhea2ecocyc", None),
+        ("kegg.reaction", "rhea2kegg_reaction", None),
+        ("reactome", "rhea2reactome", None),
+        ("macie", "rhea2macie", None),
+        ("metacyc", "rhea2metacyc", None),
+        ("go", "rhea2go", reaction_enabled_by_molecular_function),
+        ("uniprot", "rhea2uniprot", enabled_by),
     ]:
         xref_df = ensure_df(
             PREFIX,
             url=f"ftp://ftp.expasy.org/databases/rhea/tsv/{url}.tsv",
             version=version,
             force=force,
         )
-        for rhea_id, _, _, xref_id in xref_df.values:
-            if rhea_id not in terms:
+        for directional_rhea_id, _direction, _master_rhea_id, xref_id in xref_df.values:
+            if directional_rhea_id not in terms:
                 logger.debug(
                     "[%s] could not find %s:%s for xref %s:%s",
                     PREFIX,
                     PREFIX,
-                    rhea_id,
+                    directional_rhea_id,
                     xref_prefix,
                     xref_id,
                 )
                 continue
-            terms[rhea_id].append_xref(Reference(prefix=xref_prefix, identifier=xref_id))
+            target_reference = Reference(prefix=xref_prefix, identifier=xref_id)
+            if isinstance(relation, TypeDef):
+                terms[directional_rhea_id].append_relationship(relation, target_reference)
+            else:
+                terms[directional_rhea_id].append_xref(target_reference)
 
-    # TODO are EC codes equivalent?
-    # TODO uniprot enabled by (RO:0002333)
-    # TODO names?
-    # TODO participants?
+    ec_df = ensure_df(
+        PREFIX,
+        url="ftp://ftp.expasy.org/databases/rhea/tsv/rhea-ec-iubmb.tsv",
+        version=version,
+        force=force,
+    )
+    for (
+        directional_rhea_id,
+        _status,
+        _direction,
+        _master_id,
+        ec,
+        _enzyme_status,
+        _iubmb,
+    ) in ec_df.values:
+        terms[directional_rhea_id].append_relationship(
+            enabled_by, Reference(prefix="eccode", identifier=ec)
+        )
 
     yield from terms.values()
 
 
 if __name__ == "__main__":
-    RheaGetter.cli()
+    RheaGetter().write_default(write_obo=True, force=True)

src/pyobo/struct/struct.py

@@ -395,14 +395,14 @@ def _definition_fp(self) -> str:
 
     def iterate_relations(self) -> Iterable[Tuple[TypeDef, Reference]]:
         """Iterate over pairs of typedefs and targets."""
-        for typedef, targets in self.relationships.items():
-            for target in targets:
+        for typedef, targets in sorted(self.relationships.items(), key=_sort_relations):
+            for target in sorted(targets, key=lambda ref: ref.preferred_curie):
                 yield typedef, target
 
     def iterate_properties(self) -> Iterable[Tuple[str, str]]:
         """Iterate over pairs of property and values."""
-        for prop, values in self.properties.items():
-            for value in values:
+        for prop, values in sorted(self.properties.items()):
+            for value in sorted(values):
                 yield prop, value
 
     def iterate_obo_lines(self, *, ontology, typedefs) -> Iterable[str]:
@@ -466,7 +466,7 @@ def _escape(s) -> str:
 
 def _sort_relations(r):
     typedef, _references = r
-    return typedef.reference.name or typedef.reference.identifier
+    return typedef.preferred_curie
 
 
 def _sort_properties(r):

src/pyobo/struct/typedef.py

@@ -48,6 +48,10 @@
 ]
 
 
+def _bool_to_obo(v: bool) -> str:
+    return "true" if v else "false"
+
+
 @dataclass
 class TypeDef(Referenced):
     """A type definition in OBO.
@@ -88,7 +92,7 @@ def iterate_obo_lines(self) -> Iterable[str]:
             yield f'def: "{self.definition}"'
 
         if self.is_metadata_tag is not None:
-            yield f'is_metadata_tag: {"true" if self.is_metadata_tag else "false"}'
+            yield f"is_metadata_tag: {_bool_to_obo(self.is_metadata_tag)}"
 
         if self.namespace:
             yield f"namespace: {self.namespace}"
@@ -113,6 +117,10 @@ def iterate_obo_lines(self) -> Iterable[str]:
             yield f"holds_over_chain: {_chain} ! {_names}"
         if self.inverse:
             yield f"inverse_of: {self.inverse}"
+        if self.domain:
+            yield f"domain: {self.domain}"
+        if self.range:
+            yield f"range: {self.range}"
 
     @classmethod
     def from_triple(cls, prefix: str, identifier: str, name: Optional[str] = None) -> "TypeDef":
@@ -161,13 +169,19 @@ def get_reference_tuple(relation: RelationHint) -> Tuple[str, str]:
     "species with RO:0002162 (in taxon)",
 )
 has_left_to_right_reaction = TypeDef(
-    Reference(prefix="debio", identifier="0000007", name="has left-to-right reaction")
+    Reference(prefix="debio", identifier="0000007", name="has left-to-right reaction"),
+    is_metadata_tag=True,
 )
 has_right_to_left_reaction = TypeDef(
-    Reference(prefix="debio", identifier="0000008", name="has right-to-left reaction")
+    Reference(prefix="debio", identifier="0000008", name="has right-to-left reaction"),
+    is_metadata_tag=True,
 )
 has_bidirectional_reaction = TypeDef(
-    Reference(prefix="debio", identifier="0000009", name="has bi-directional reaction")
+    Reference(prefix="debio", identifier="0000009", name="has bi-directional reaction"),
+    is_metadata_tag=True,
+)
+reaction_enabled_by_molecular_function = TypeDef(
+    Reference(prefix="debio", identifier="0000047", name="reaction enabled by molecular function")
 )
 
 
@@ -291,7 +305,14 @@ def get_reference_tuple(relation: RelationHint) -> Tuple[str, str]:
 is_immediately_transformed_from = TypeDef.from_triple(
     prefix=SIO_PREFIX, identifier="000658", name="is immediately transformed from"
 )
-enables = TypeDef.from_triple(prefix="RO", identifier="0002327", name="enables")
+
+_enables_reference = Reference(prefix=RO_PREFIX, identifier="0002327", name="enables")
+_enabled_by_reference = Reference(prefix=RO_PREFIX, identifier="0002333", name="enabled by")
+enables = TypeDef(reference=_enables_reference, inverse=_enabled_by_reference)
+enabled_by = TypeDef(reference=_enabled_by_reference, inverse=_enables_reference)
+
+has_input = TypeDef.from_triple(prefix=RO_PREFIX, identifier="0002233", name="has input")
+has_output = TypeDef.from_triple(prefix=RO_PREFIX, identifier="0002234", name="has output")
 
 """ChEBI"""