adding Clinical Connections, cleaning up

RobokopU24 · Oct 28, 2024 · 7005505 · 7005505
1 parent 5c42ff8
commit 7005505
Show file tree

Hide file tree

Showing 2 changed files with 27 additions and 4 deletions.
diff --git a/Common/data_sources.py b/Common/data_sources.py
@@ -9,6 +9,8 @@
 CTD = 'CTD'
 DRUG_CENTRAL = 'DrugCentral'
 DRUGMECHDB = 'DrugMechDB'
+EHRMAYTREAT = 'EHRMayTreat'
+EHRCLINICALCONNECTIONS = 'EHRClinicalConnections'
 # FOODB = 'FooDB' # this is on hold, data needs review after latest release of data.
 GENOME_ALLIANCE_ORTHOLOGS = 'GenomeAllianceOrthologs'
 GTEX = 'GTEx'
@@ -23,7 +25,6 @@
 LITCOIN_SAPBERT = 'LitCoinSapBERT'
 LITCOIN_ENTITY_EXTRACTOR = 'LitCoinEntityExtractor'
 KINACE = 'KinAce'
-MAYTREATKP = 'EHRMayTreatKP'
 MOLEPRO = 'MolePro'
 MONARCH_KG = 'MonarchKG'
 MONDO_PROPS = 'MONDOProps'
@@ -72,7 +73,8 @@
     LITCOIN_ENTITY_EXTRACTOR: ("parsers.LitCoin.src.loadLitCoin", "LitCoinEntityExtractorLoader"),
     LITCOIN_SAPBERT: ("parsers.LitCoin.src.loadLitCoin", "LitCoinSapBERTLoader"),
     KINACE: ("parsers.KinAce.src.loadKinAce", "KinAceLoader"),
-    MAYTREATKP: ("parsers.ehr.src.loadEHR", "EHRMayTreatLoader"),
+    EHRMAYTREAT: ("parsers.ehr.src.loadEHR", "EHRMayTreatLoader"),
+    EHRCLINICALCONNECTIONS: ("parsers.ehr.src.loadEHR", "EHRClinicalConnectionsLoader"),
     MOLEPRO: ("parsers.molepro.src.loadMolePro", "MoleProLoader"),
     MONARCH_KG: ("parsers.monarchkg.src.loadMonarchKG", "MonarchKGLoader"),
     MONDO_PROPS: ("parsers.MONDOProperties.src.loadMP", "MPLoader"),

diff --git a/parsers/ehr/src/loadEHR.py b/parsers/ehr/src/loadEHR.py
@@ -24,8 +24,8 @@ def __init__(self, test_mode: bool = False, source_data_dir: str = None):
         """
         super().__init__(test_mode=test_mode, source_data_dir=source_data_dir)
 
-        self.data_url: str = 'https://stars.renci.org/var/data_services/ehrmaytreatkp/'
-        self.edge_file_name: str = 'edges.csv'
+        self.data_url: str = 'https://raw.githubusercontent.com/Hadlock-Lab/EHRMLA_KPs/main/May%20Treat/csvs/'
+        self.edge_file_name: str = 'May_Treat_KP_edges_latest_deploy.csv'
         self.data_file = self.edge_file_name
         self.version_file = 'ehr-may-treat-kp.yaml'
 
@@ -86,3 +86,24 @@ def parse_data(self) -> dict:
             'unusable_source_lines': skipped_record_counter}
         return load_metadata
 
+
+class EHRClinicalConnectionsLoader(EHRMayTreatLoader):
+    source_id: str = "EHRClinicalConnectionsKP"
+    provenance_id: str = "infores:isb-EHRMLA-clinicalconnections"
+    description = "Multiomics EHRMLA Clinical Connections KP."
+    source_data_url = "https://github.com/NCATSTranslator/Translator-All/wiki/Multiomics-EHRMLA-May-Treat-KP"
+    license = "https://github.com/NCATSTranslator/Translator-All/wiki/Multiomics-EHRMLA-May-Treat-KP"
+    attribution = "https://github.com/NCATSTranslator/Translator-All/wiki/Multiomics-EHRMLA-May-Treat-KP"
+    parsing_version = "1.0"
+
+    def __init__(self, test_mode: bool = False, source_data_dir: str = None):
+        """
+        :param test_mode - sets the run into test mode
+        :param source_data_dir - the specific storage directory to save files in
+        """
+        super().__init__(test_mode=test_mode, source_data_dir=source_data_dir)
+
+        self.data_url: str = 'https://raw.githubusercontent.com/Hadlock-Lab/EHRMLA_KPs/main/Clinical%20Connections/csvs/'
+        self.edge_file_name: str = 'ClinicalConnections_KP_edges_latest_deploy.csv'
+        self.data_file = self.edge_file_name
+        self.version_file = 'ehr-clinical-connections-kp.yaml'