Adding uniprot adapter. (#319)

* Adding uniprot adapter.

This is still preliminary.

* Fixing mapping test

src/oaklib/cli.py

@@ -578,6 +578,7 @@ def chain_it(v):
         else:
             if isinstance(impl, SearchInterface):
                 cfg = create_search_configuration(term)
+                logging.info(f"Search config: {term} => {cfg}")
                 chain_it(impl.basic_search(cfg.search_terms[0], config=cfg))
             else:
                 raise NotImplementedError
@@ -2235,12 +2236,7 @@ def relationships(
 
     """
     impl = settings.impl
-    if output_type == "obo":
-        writer = StreamingOboWriter(ontology_interface=impl)
-    elif output_type == "csv":
-        writer = StreamingCsvWriter(ontology_interface=impl)
-    else:
-        writer = StreamingCsvWriter(ontology_interface=impl)
+    writer = _get_writer(output_type, impl, StreamingCsvWriter)
     writer.autolabel = autolabel
     writer.output = output
     actual_predicates = _process_predicates_arg(predicates)
@@ -3683,6 +3679,7 @@ def diff_via_mappings(
         logging.info("No term list provided, will compare all mapped terms")
         entities = None
     actual_predicates = _process_predicates_arg(predicates)
+    n = 0
     for r in calculate_pairwise_relational_diff(
         oi,
         other_oi,
@@ -3696,6 +3693,9 @@ def diff_via_mappings(
         if filter_category_identical and r.category == DiffCategory(DiffCategory.Identical):
             continue
         writer.emit(r)
+        n += 1
+    if n == 0:
+        raise ValueError("No mappings extracted")
 
 
 @main.command()

src/oaklib/datamodels/search.py

@@ -8,7 +8,11 @@
     SearchProperty,
     SearchTermSyntax,
 )
-from oaklib.datamodels.vocabulary import LABEL_PREDICATE, SYNONYM_PREDICATES
+from oaklib.datamodels.vocabulary import (
+    IDENTIFIER_PREDICATE,
+    LABEL_PREDICATE,
+    SYNONYM_PREDICATES,
+)
 from oaklib.types import PRED_CURIE
 
 DEFAULT_SEARCH_PROPERTIES = [SearchProperty.LABEL, SearchProperty.ALIAS]
@@ -68,6 +72,8 @@ def search_properties_to_predicates(props: List[SearchProperty]) -> List[PRED_CU
     for p in props:
         if p == SearchProperty(SearchProperty.LABEL):
             preds.add(LABEL_PREDICATE)
+        elif p == SearchProperty(SearchProperty.IDENTIFIER):
+            preds.add(IDENTIFIER_PREDICATE)
         elif p == SearchProperty(SearchProperty.ALIAS):
             preds.update(SYNONYM_PREDICATES + [LABEL_PREDICATE])
         elif p == SearchProperty(SearchProperty.ANYTHING):

src/oaklib/datamodels/search_datamodel.yaml

@@ -1,11 +1,17 @@
 id: https://w3id.org/linkml/search_datamodel
 title: search datamodel
 name: search-datamodel
-description: >-
-  A datamodel for representing a search configuration and results. This is intended to provide a unified layer over
-  both (a) how searches are *parameterized* (b) the structure of search *results*. The scope is any kind of service
-  that provides search over *named entities*, including ontology concepts. It is not intended to cover generic
+description: |-
+  A datamodel for representing a search configuration and results.
+  
+  This is intended to provide a unified layer over both:
+  
+  - (a) how searches are *parameterized*
+  - (b) the structure of search *results*.
+  
+  The scope is any kind of service that provides search over *named entities*, including ontology concepts. It is not intended to cover generic
   search results, e.g. google search, although parts could be generalized for this purpose.
+
 license: https://creativecommons.org/publicdomain/zero/1.0/
 
 prefixes:

src/oaklib/datamodels/vocabulary.py

@@ -12,6 +12,8 @@
 
 APP_NAME = "ontology-access-kit"
 
+IDENTIFIER_PREDICATE = "rdf:ID"
+
 # TODO: replace with oio vocab
 LABEL_PREDICATE = omd.slots.label.curie
 HAS_EXACT_SYNONYM = omd.slots.has_exact_synonym.curie
@@ -51,6 +53,7 @@
 RDF_TYPE = "rdf:type"
 EQUIVALENT_CLASS = "owl:equivalentClass"
 OWL_SAME_AS = "owl:sameAs"
+RDF_SEE_ALSO = "rdfs:seeAlso"
 PART_OF = "BFO:0000050"
 LOCATED_IN = "RO:0001025"
 DEVELOPS_FROM = "RO:0002202"

src/oaklib/implementations/sparql/abstract_sparql_implementation.py

@@ -23,11 +23,12 @@
     SearchConfiguration,
     search_properties_to_predicates,
 )
-from oaklib.datamodels.search_datamodel import SearchProperty, SearchTermSyntax
+from oaklib.datamodels.search_datamodel import SearchTermSyntax
 from oaklib.datamodels.vocabulary import (
     ALL_MATCH_PREDICATES,
     DEFAULT_PREFIX_MAP,
     HAS_DEFINITION_URI,
+    IDENTIFIER_PREDICATE,
     IS_A,
     LABEL_PREDICATE,
     OBO_PURL,
@@ -162,6 +163,14 @@ def _label_uri(self):
             else RDFS.label
         )
 
+    def _mapping_predicates(self):
+        preds = ALL_MATCH_PREDICATES
+        omm = self.ontology_metamodel_mapper
+        if omm:
+            return [omm.map_curie(pred, unmapped_reflexive=True) for pred in preds]
+        else:
+            return preds
+
     def _definition_uri(self):
         return (
             self.ontology_metamodel_mapper.definition_uri()
@@ -221,9 +230,10 @@ def obsoletes(self) -> Iterable[CURIE]:
             yield self.uri_to_curie(row["s"]["value"])
 
     def simple_mappings_by_curie(self, curie: CURIE) -> Iterable[Tuple[PRED_CURIE, CURIE]]:
+        mapping_preds = self._mapping_predicates()
         uri = self.curie_to_sparql(curie)
         query = SparqlQuery(select=["?p ?o"], distinct=True, where=[f"{uri} ?p ?o"])
-        query.add_values("p", [self.curie_to_sparql(p) for p in ALL_MATCH_PREDICATES])
+        query.add_values("p", [self.curie_to_sparql(p) for p in mapping_preds])
         bindings = self._query(query.query_str())
         for row in bindings:
             yield (self.uri_to_curie(row["p"]["value"]), self.uri_to_curie(row["o"]["value"]))
@@ -243,7 +253,7 @@ def list_of_named_graphs(self) -> List[URI]:
         sw.setReturnFormat(JSON)
         check_limit()
         ret = sw.queryAndConvert()
-        logging.info(f"RET={ret}")
+        logging.debug(f"RET={ret}")
         self._list_of_named_graphs = [row["g"]["value"] for row in ret["results"]["bindings"]]
         return self._list_of_named_graphs
 
@@ -290,7 +300,7 @@ def tr(v: Identifier):
             sw.setReturnFormat(JSON)
             check_limit()
             ret = sw.queryAndConvert()
-            logging.info(f"RET={ret}")
+            logging.debug(f"queryResults={ret}")
             return ret["results"]["bindings"]
 
     def _triples(
@@ -605,8 +615,9 @@ def basic_search(
         self, search_term: str, config: SearchConfiguration = SEARCH_CONFIG
     ) -> Iterable[CURIE]:
         if ":" in search_term and " " not in search_term:
-            logging.debug(f"Not performing search on what looks like a CURIE: {search_term}")
-            return
+            # logging.error(f"Not performing search on what looks like a CURIE: {search_term}")
+            # return
+            search_term = self.curie_to_uri(search_term)
 
         if self._is_blazegraph():
             filter_clause = f'?v bds:search "{search_term}"'
@@ -631,18 +642,29 @@ def basic_search(
                 self.ontology_metamodel_mapper.map_curie(pred, unmapped_reflexive=True)[0]
                 for pred in preds
             ]
-        preds = [self.curie_to_sparql(p) for p in preds]
-        if len(preds) == 1:
-            where = [f"?s {preds[0]} ?v "]
+        if preds == [IDENTIFIER_PREDICATE]:
+            where = ["?v a ?s_cls", "BIND(?v AS ?s)"]
+            query = SparqlQuery(select=["?s"], where=where + [filter_clause])
         else:
-            where = ["?s ?p ?v ", f'VALUES ?p {{ {" ".join(preds)} }}']
-        query = SparqlQuery(select=["?s"], where=where + [filter_clause])
-        # print(f"Search query: {query.query_str()}")
+            non_id_preds = [pred for pred in preds if pred != IDENTIFIER_PREDICATE]
+            non_id_preds = [self.curie_to_sparql(p) for p in non_id_preds]
+            if len(non_id_preds) == 1:
+                where = [f"?s {preds[0]} ?v "]
+            elif len(non_id_preds) == 1:
+                raise ValueError("Logic error; this should be handled by above clause")
+            else:
+                where = ["?s ?p ?v ", f'VALUES ?p {{ {" ".join(non_id_preds)} }}']
+            if IDENTIFIER_PREDICATE in preds:
+                raise NotImplementedError(
+                    f"Cannot mix identifier and non-identifier preds: {preds}"
+                )
+            query = SparqlQuery(select=["?s"], where=where + [filter_clause])
+        logging.info(f"Search query: {query.query_str()}")
         bindings = self._query(query, prefixes=DEFAULT_PREFIX_MAP)
         for row in bindings:
             yield self.uri_to_curie(row["s"]["value"])
-        if SearchProperty(SearchProperty.IDENTIFIER) in config.properties:
-            raise NotImplementedError
+        # if SearchProperty(SearchProperty.IDENTIFIER) in config.properties:
+        #    raise NotImplementedError
 
     # ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
     # Implements: OboGraphInterface
@@ -665,7 +687,7 @@ def hierarchical_descendants(self, start_curies: Union[CURIE, List[CURIE]]) -> I
     # ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 
     def get_sssom_mappings_by_curie(self, curie: CURIE) -> Iterable[Mapping]:
-        pred_uris = [self.curie_to_sparql(pred) for pred in ALL_MATCH_PREDICATES]
+        pred_uris = [self.curie_to_sparql(pred) for pred in self._mapping_predicates()]
         # input curie is subject
         query = SparqlQuery(
             select=["?p", "?o"],

src/oaklib/implementations/ubergraph/ubergraph_implementation.py

@@ -66,7 +66,7 @@ class UbergraphImplementation(
 
         .. code:: python
 
-           >>>  oi = UbergraphImplementation.create()
+           >>>  oi = UbergraphImplementation()
 
         The default ubergraph endpoint will be assumed