Gocamgen reports: Logical def errors and GO rules

bin/validate.py

@@ -566,9 +566,11 @@ def gpad2gocams(ctx, gpad_path, gpi_path, target, ontology, ttl):
     gpad_basename = os.path.basename(gpad_path)
     gpad_basename_root, gpad_ext = os.path.splitext(gpad_basename)
     output_basename = "{}.nq".format(gpad_basename_root)
-    report_basename = "{}.gocamgen.report".format(gpad_basename_root)
+    parse_report_basename = "{}.parser.report".format(gpad_basename_root)
+    model_report_basename = "{}.gocamgen.report".format(gpad_basename_root)
     output_path = os.path.join(absolute_target, output_basename)
-    report_path = os.path.join(absolute_target, report_basename)
+    parse_report_path = os.path.join(absolute_target, parse_report_basename)
+    model_report_path = os.path.join(absolute_target, model_report_basename)
 
     builder = GoCamBuilder(parser_config=parser_config)
 
@@ -580,7 +582,10 @@ def gpad2gocams(ctx, gpad_path, gpi_path, target, ontology, ttl):
     if not ttl:
         builder.write_out_store_to_nquads(filepath=output_path)
 
-    builder.write_report(report_filepath=report_path)
+    # Reports
+    builder.write_report(report_filepath=model_report_path)
+    with open(parse_report_path, 'w') as prp:
+        prp.write(extractor.gpad_parser.report.to_markdown())
 
 
 @cli.command()

ontobio/rdfgen/gocamgen/errors.py

@@ -1,5 +1,8 @@
+from ontobio.rdfgen.gocamgen import collapsed_assoc
+
+
 class GocamgenException(Exception):
-    def __init__(self, message):
+    def __init__(self, message: str):
         self.message = message
 
     def __str__(self):
@@ -14,6 +17,15 @@ class ShexException(GocamgenException):
     pass
 
 
+class CollapsedAssocGocamgenException(GocamgenException):
+    def __init__(self, message: str, assoc: collapsed_assoc.CollapsedAssociation):
+        self.message = message
+        self.assoc = assoc
+
+    def __str__(self):
+        return "{}\n{}".format(self.message, "\n".join([l.source_line for l in self.assoc.lines]))
+
+
 class GeneErrorSet:
     def __init__(self):
         self.errors = {}

ontobio/rdfgen/gocamgen/gocam_builder.py

@@ -76,6 +76,8 @@ def make_model(self, gene, annotations: List[GoAssociation], output_directory=No
                     model.write(out_filename)
                     logger.info("Model for {} written to {} in {} sec".format(gene, out_filename,
                                                                               (time.time() - start_time)))
+                for err in model.errors:
+                    self.errors.add_error(gene, err)
                 return model
             except GocamgenException as ex:
                 self.errors.add_error(gene, ex)
@@ -156,13 +158,13 @@ def parse_gpi(gpi_file):
 class AssocExtractor:
     def __init__(self, gpad_file, parser_config: AssocParserConfig):
         self.assocs = []
-        gpad_parser = gpadparser.GpadParser(config=parser_config)
+        self.gpad_parser = gpadparser.GpadParser(config=parser_config)
         with open(gpad_file) as sg:
             lines = sum(1 for line in sg)
 
         with open(gpad_file) as gf:
             click.echo("Making products...")
-            with click.progressbar(iterable=gpad_parser.association_generator(file=gf, skipheader=True),
+            with click.progressbar(iterable=self.gpad_parser.association_generator(file=gf, skipheader=True),
                                    length=lines) as associations:
                 self.assocs = list(associations)
 

ontobio/rdfgen/gocamgen/gocamgen.py

@@ -463,6 +463,7 @@ def __init__(self, modeltitle, assocs: List[GoAssociation], config: AssocParserC
         self.provided_bys = set()
         self.graph.bind("GOREL", GOREL)  # Because GOREL isn't in context.jsonld's
         self.gpi_entities = gpi_entities
+        self.errors: List[errors.GocamgenException] = []
         ncbi_taxon = self.taxon_id_from_entity(str(assocs[0].subject.id))
         # Emit model-level in_taxon triple from ncbi_taxon
         if ncbi_taxon:
@@ -615,7 +616,9 @@ def translate(self):
                                     ext_target_n = annot_subgraph.add_instance_of_class(ext_target)
                                     annot_subgraph.add_edge(regulated_term_n, chained_rel, ext_target_n)
                                 else:
-                                    logger.warning("Couldn't get regulates relation from LD of: {}".format(term))
+                                    err_msg = "Couldn't get regulates relation from LD of: {}".format(term)
+                                    logger.warning(err_msg)
+                                    self.errors.append(errors.CollapsedAssocGocamgenException(err_msg, a))
                             elif ext_relation in HAS_REGULATION_TARGET_RELATIONS:
                                 if aspect == 'P':
                                     # For BP annotations, translate 'has regulation target' to 'has input'.