Check with-from ID taxon in GPI - geneontology/gocamgen#79; pipe spli…

…tting for geneontology/gocamgen#84

ontobio/rdfgen/gocamgen/collapsed_assoc.py

@@ -14,6 +14,7 @@ def __init__(self, associations):
         self.collapsed_associations = []
         self.assoc_dict = {}
         self.go_ontology = None
+        self.gpi_entities = None
 
     def setup_ontologies(self):
         if self.go_ontology is None:
@@ -43,20 +44,14 @@ def collapse_annotations(self):
             subj_id = a["subject"]["id"]
             qualifiers = a["qualifiers"]
             term = a["object"]["id"]
-            with_from = a["evidence"]["with_support_from"]
-            eco_code = a["evidence"]["type"]
             extensions = get_annot_extensions(a)
-            with_froms = get_with_froms(a)  # Handle pipe separation according to import requirements
-            is_protein_binding = eco_code == IPI_ECO_CODE and BINDING_ROOT in self.go_ontology.ancestors(term, reflexive=True)
-            if is_protein_binding:
-                cas = self.find_or_create_collapsed_associations(subj_id, qualifiers, term, with_froms, extensions)
-                with_from = None  # Don't use ontobio-parsed with_from on lines
-            else:
-                cas = [self.find_or_create_collapsed_association(subj_id, qualifiers, term, None, extensions)]
-            for ca in cas:
-                # Line
-                association_line = CollapsedAssociationLine(a, with_from)
+            with_froms = self.get_with_froms(a)  # Handle pipe separation according to import requirements
+            cas = []
+            for wf in with_froms:
+                ca = self.find_or_create_collapsed_association(subj_id, qualifiers, term, wf, extensions)
+                association_line = CollapsedAssociationLine(a, wf["line"])
                 ca.lines.append(association_line)
+                cas.append(ca)
 
     def find_or_create_collapsed_association(self, subj_id, qualifiers, term, with_from, extensions):
         query_header = {
@@ -69,25 +64,61 @@ def find_or_create_collapsed_association(self, subj_id, qualifiers, term, with_f
             },
             'object_extensions': extensions
         }
-        if with_from:
-            query_header['evidence'] = {'with_support_from': sorted(with_from)}
+        if with_from and "header" in with_from:
+            query_header['evidence'] = {'with_support_from': sorted(with_from['header'])}
         for ca in self.collapsed_associations:
             if ca.header == query_header:
                 return ca
         new_ca = CollapsedAssociation(query_header)
         self.collapsed_associations.append(new_ca)
         return new_ca
 
-    def find_or_create_collapsed_associations(self, subj_id, qualifiers, term, with_froms, extensions):
-        cas = []
-        for wf in with_froms:
-            ca = self.find_or_create_collapsed_association(subj_id, qualifiers, term, wf, extensions)
-            cas.append(ca)
-        return cas
-
     def __iter__(self):
         return iter(self.collapsed_associations)
 
+    def get_with_froms(self, annot):
+        source_line = annot["source_line"]
+        vals = source_line.split("\t")
+        with_from_col = vals[6]
+        # Parse into array (by "|") of arrays (by ",")
+        with_from_ds = []  # The list of lists
+        for piped_with_from in with_from_col.split("|"):
+            # Will be bypassing ontobio ID validation? Let's try teaming up with ontobio functions!
+            split_line = SplitLine(line=source_line, values=vals, taxon="")  # req'd for error reporting in ontobio?
+            validated_comma_with_froms = GPAD_PARSER.validate_pipe_separated_ids(piped_with_from, split_line,
+                                                                                 empty_allowed=True, extra_delims=",")
+            with_from_ds.append(validated_comma_with_froms)
+
+        # Now arrange these into "header" and "line" values
+        eco_code = annot["evidence"]["type"]
+        term = annot["object"]["id"]
+        is_binding = eco_code == IPI_ECO_CODE and BINDING_ROOT in self.go_ontology.ancestors(term, reflexive=True)
+        if is_binding:
+            # Using GPI, check with_froms for taxon equivalency to subj_id
+            if self.gpi_entities:
+                subject_id = annot["subject"]["id"]
+                subject_entity = self.gpi_entities[subject_id]
+                values_separated = []
+                for wf in with_from_ds:
+                    wf_separated = {
+                        "header": [],
+                        "line": []
+                    }
+                    for wf_id in wf:
+                        wf_entity = self.gpi_entities.get(wf_id)
+                        if wf_entity and wf_entity.get("taxon") == subject_entity["taxon"]:
+                            wf_separated['header'].append(wf_id)
+                        else:
+                            wf_separated['line'].append(wf_id)
+                    values_separated.append(wf_separated)
+                return values_separated
+            else:
+                # Everything is defaulted to header if no GPI available
+                return [{"header": wf} for wf in with_from_ds]
+        else:
+            # Everything is defaulted to line if not binding
+            return [{"line": wf} for wf in with_from_ds]
+
 
 class CollapsedAssociation:
     def __init__(self, header):
@@ -168,23 +199,6 @@ def get_annot_extensions(annot):
     return {}
 
 
-def get_with_froms(annot):
-    source_line = annot["source_line"]
-    vals = source_line.split("\t")
-    with_from_col = vals[6]
-    # Parse into array (by "|") of arrays (by ",")
-    with_from_ds = []
-    for piped_with_from in with_from_col.split("|"):
-        # Will be bypassing ontobio ID validation? Let's try teaming up with ontobio functions!
-        split_line = SplitLine(line=source_line, values=vals, taxon="")  # req'd for error reporting in ontobio?
-        validated_comma_with_froms = GPAD_PARSER.validate_pipe_separated_ids(piped_with_from, split_line, empty_allowed=True, extra_delims=",")
-        # comma_with_froms = piped_with_from.split(",")
-        # validated_comma_with_froms = []
-        # for wf in comma_with_froms:
-        with_from_ds.append(validated_comma_with_froms)
-    return with_from_ds
-
-
 def extract_properties_from_string(prop_col):
     props = prop_col.split("|")
     props_dict = {}