Merge pull request #541 from biolink/pipeline-206-small-updates

updates needed for new pipeline kernel, small bugifxes

ontobio/io/assocparser.py

@@ -21,7 +21,7 @@
 from dataclasses import dataclass
 
 from collections import namedtuple, defaultdict
-from typing import Optional, List, Dict, Set
+from typing import Optional, List, Dict, Set, Union
 
 from ontobio import ontol
 from ontobio import ecomap
@@ -897,9 +897,6 @@ def parse_with_dateutil(date: str, repot: Report, line: List) -> Optional[associ
 
     return d
 
-def parse_annotation_properties(properties_field: str):
-    properties_list = [tuple(prop.split("=", maxsplit=1)) for prop in properties_field.split("|") if prop]
-    return properties_list
 
 ## we generate both qualifier and relation field
 ## Returns: (negated, relation, other_qualifiers)

ontobio/io/gaference.py

@@ -10,6 +10,7 @@
 from ontobio.model import association
 from ontobio.rdfgen import relations
 
+import functools
 import logging
 import copy
 
@@ -84,6 +85,9 @@ def load_gaferencer_inferences_from_file(gaferencer_out) -> Dict[AnnotationKey,
 
     return build_annotation_inferences(gaferencer_out_dict)
 
+def load_gaferencer_inferences_from_files(gaferencer_out_list) -> Dict[AnnotationKey, InferenceValue]:
+    return functools.reduce(load_gaferencer_inferences_from_file, gaferencer_out_list, dict())
+
 
 def build_annotation_inferences(gaferencer_out: List[Dict]) -> Dict[AnnotationKey, InferenceValue]:
     inferences = dict()  # type Dict[AnnotationKey, InferenceValue]

ontobio/io/gpadparser.py

@@ -330,7 +330,7 @@ def from_1_2(gpad_line: List[str], report=None, group="unknown", dataset="unknow
             report.error(source_line, Report.EXTENSION_SYNTAX_ERROR, conjunctions.info, "extensions should be relation(curie)", taxon=str(taxon), rule=1)
             return assocparser.ParseResult(source_line, [], True, report=report)
 
-    properties_list = assocparser.parse_annotation_properties(gpad_line[11])
+    properties_list = association.parse_annotation_properties(gpad_line[11])
 
 
     # print(properties_list)
@@ -456,7 +456,7 @@ def from_2_0(gpad_line: List[str], report=None, group="unknown", dataset="unknow
             report.error(source_line, Report.EXTENSION_SYNTAX_ERROR, conjunctions.info, "extensions should be relation(curie)", taxon=str(taxon), rule=1)
             return assocparser.ParseResult(source_line, [], True, report=report)
 
-    properties_list = assocparser.parse_annotation_properties(gpad_line[11])
+    properties_list = association.parse_annotation_properties(gpad_line[11])
 
     a = association.GoAssociation(
         source_line=source_line,

ontobio/model/association.py

@@ -30,6 +30,7 @@
 def ymd_str(date: Date, separator: str) -> str:
     return "{year}{sep}{month}{sep}{day}".format(year=date.year, sep=separator, month=date.month, day=date.day)
 
+
 @dataclass
 class Error:
     info: str
@@ -38,6 +39,27 @@ class Error:
     def is_error(self):
         return True
 
+def TwoTupleStr(items: List[str]) -> tuple:
+    """
+    Create a tuple of of str that is guaranteed to be of length two from a list
+
+    If the list is larger, then only the first two elements will be used.
+    If the list is smaller, then the empty string will be used
+    """
+
+    if len(items) > 2:
+        items = items[:2]
+
+    if len(items) < 2:
+        items += [""] * (2 - len(items))
+
+    return tuple([str(i) for i in items])
+
+
+def parse_annotation_properties(properties_field: str) -> List[tuple]:
+    properties_list = [TwoTupleStr(prop.split("=", maxsplit=1)) for prop in properties_field.split("|") if prop]
+    return properties_list
+
 @dataclass(unsafe_hash=True)
 class Curie:
     namespace: str

tests/test_gpad_parser.py

@@ -1,5 +1,6 @@
 from ontobio.io.gpadparser import GpadParser, to_association
 from ontobio.io import assocparser
+from ontobio.model import association
 from ontobio.model.association import ConjunctiveSet, ExtensionUnit, Curie
 
 import yaml
@@ -61,7 +62,7 @@ def test_parse_interacting_taxon():
 
 def test_duplicate_key_annot_properties():
     properties_str = "creation-date=2008-02-07|modification-date=2010-12-01|comment=v-KIND domain binding of Kndc1;MGI:1923734|contributor-id=http://orcid.org/0000-0003-2689-5511|contributor-id=http://orcid.org/0000-0003-3394-9805"
-    prop_list = assocparser.parse_annotation_properties(properties_str)
+    prop_list = association.parse_annotation_properties(properties_str)
     contributor_ids = [value for key, value in prop_list if key == "contributor-id"]
     assert set(contributor_ids) == {"http://orcid.org/0000-0003-2689-5511", "http://orcid.org/0000-0003-3394-9805"}