Merge pull request #398 from apriltuesday/issue-393

Add record creation date to evidence

OT_SCHEMA_VERSION

@@ -1 +1 @@
-2.4.0
+022fabb

bin/trait_mapping/create_table_for_manual_curation.py

@@ -14,8 +14,7 @@ def previous_and_replacement_mappings(trait_name, previous_mappings):
     if trait_name not in previous_mappings:
         yield '', ''
         return
-    for uri in previous_mappings[trait_name]:
-        label = get_ontology_label(uri)
+    for uri, label in previous_mappings[trait_name]:
         trait_status = get_trait_status(uri)
         trait_string = '|'.join([uri, label, 'NOT_SPECIFIED', 'previously-used', trait_status])
         replacement_string = find_replacement_mapping(uri)

cmat/clinvar_xml_io/clinvar_record.py

@@ -75,6 +75,11 @@ def date(self):
         """This tracks the latest update date, counting even minor technical updates."""
         return self.rcv.attrib['DateLastUpdated']
 
+    @property
+    def created_date(self):
+        """This tracks the date the record was first made public on ClinVar."""
+        return self.rcv.attrib['DateCreated']
+
     @property
     def last_evaluated_date(self):
         """This tracks the latest (re)evaluation date for the clinical interpretation.

cmat/output_generation/clinvar_to_evidence_strings.py

@@ -206,6 +206,15 @@ def clinvar_to_evidence_strings(string_to_efo_mappings, variant_to_gene_mappings
     return report
 
 
+def format_creation_date(s):
+    if not s:
+        return None
+    m = re.search('\d{4}-\d{2}-\d{2}', s)
+    if m and m.group(0):
+        return m.group(0)
+    return None
+
+
 def generate_evidence_string(clinvar_record, allele_origins, disease_name, disease_source_id, disease_mapped_efo_id,
                              consequence_attributes):
     """Generates an evidence string based on ClinVar record and some additional attributes."""
@@ -235,6 +244,9 @@ def generate_evidence_string(clinvar_record, allele_origins, disease_name, disea
         # RCV identifier.
         'studyId': clinvar_record.accession,
 
+        # Record creation date, formatted as YYYY-MM-DD
+        'releaseDate': format_creation_date(clinvar_record.created_date),
+
         # VARIANT ATTRIBUTES.
         'targetFromSourceId': consequence_attributes.ensembl_gene_id,
         'variantFunctionalConsequenceId': consequence_attributes.so_term.accession,

data-exploration/filter_clinvar_xml.py

@@ -3,10 +3,10 @@
 import logging
 import xml.etree.ElementTree as ElementTree
 
-from eva_cttv_pipeline.clinvar_xml_io.clinvar_xml_io import ClinVarRecord, find_mandatory_unique_element
-from eva_cttv_pipeline.evidence_string_generation.clinvar_to_evidence_strings import get_consequence_types
-from eva_cttv_pipeline.evidence_string_generation.consequence_type import process_consequence_type_file
-
+from cmat.clinvar_xml_io import ClinVarRecord
+from cmat.clinvar_xml_io.xml_parsing import find_mandatory_unique_element
+from cmat.output_generation.clinvar_to_evidence_strings import get_consequence_types
+from cmat.output_generation.consequence_type import process_consequence_type_file
 
 logging.basicConfig()
 logger = logging.getLogger(__name__)

tests/output_generation/resources/expected_genetics_evidence_string.json

@@ -19,6 +19,7 @@
     "15258582",
     "15322982"
   ],
+  "releaseDate": "2012-08-13",
   "studyId": "RCV000002127",
   "targetFromSourceId": "ENSG00000139988",
   "variantFunctionalConsequenceId": "SO_0001583",

tests/output_generation/resources/expected_multiple_names_evidence_string.json

@@ -20,6 +20,7 @@
   "diseaseFromSource": "Skeletal dysplasia",
   "diseaseFromSourceId": "C0410528",
   "diseaseFromSourceMappedId": "HP_0002652",
+  "releaseDate": "2017-01-12",
   "studyId": "RCV000415158",
   "targetFromSourceId": "ENSG00000139988",
   "variantFunctionalConsequenceId": "SO_0001583",

tests/output_generation/resources/expected_somatic_evidence_string.json

@@ -19,6 +19,7 @@
     "15258582",
     "15322982"
   ],
+  "releaseDate": "2012-08-13",
   "studyId": "RCV000002127",
   "targetFromSourceId": "ENSG00000139988",
   "variantFunctionalConsequenceId": "SO_0001583",

tests/output_generation/test_clinvar_to_evidence_strings.py

@@ -1,5 +1,7 @@
 import json
 import os
+import re
+
 import requests
 import xml.etree.ElementTree as ElementTree