cliniwiz_keywords.diff_orig.tsv moved file

expand.sql
fixed table create "expand_keywords"

make_all.sh
now links "curated.tsv" per dataset, good enough to run

vsac.py
ported from Matts branch

cumulus_library_opioid/vocab/analyze/expand.sql

@@ -44,21 +44,18 @@ and     rules.TTY2 = relax.TTY2
 and     rules.RELA = relax.RELA ;
 
 
+-- ##############################################
+call log('expand_keywords', 'refresh');
+
 -- ##############################################
 call log('expand_keywords', 'refresh');
 
 drop    table if exists expand_keywords;
 create  table           expand_keywords
 select  distinct        relax.*
-from    expand_relax as relax,
-        expand_rules as rules
-where   rules.include is NULL
-and     relax.REL not in ('RB', 'PAR')
-and     relax.keyword_len >= 4
-and     rules.TTY1 = relax.TTY1
-and     rules.TTY2 = relax.TTY2
-and     rules.RELA = relax.RELA ;
-
+from    expand_relax as relax
+where   relax.REL not in ('RB', 'PAR')
+and     relax.keyword_len >=4;
 
 -- ##############################################
 call log('expand', 'refresh');

cumulus_library_opioid/vocab/make_all.sh

@@ -4,46 +4,58 @@ source db.config
 source env_table_schema.sh
 
 export CURATED='acep'
+cd data; rm -f curated.tsv; ln -s VSAC/$CURATED.tsv curated.tsv; cd ..
 ./make.sh
 
 export CURATED='bioportal'
+cd data; rm -f curated.tsv; ln -s bioportal.curated.tsv curated.tsv; cd ..
 ./make.sh
 
 export CURATED='CancerLinQ'
+cd data; rm -f curated.tsv; ln -s VSAC/$CURATED.tsv curated.tsv; cd ..
 ./make.sh
 
 export CURATED='ecri'
+cd data; rm -f curated.tsv; ln -s VSAC/$CURATED.tsv curated.tsv; cd ..
 ./make.sh
 
 export CURATED='impaq'
+cd data; rm -f curated.tsv; ln -s VSAC/$CURATED.tsv curated.tsv; cd ..
 ./make.sh
 
 export CURATED='lantana'
+cd data; rm -f curated.tsv; ln -s VSAC/$CURATED.tsv curated.tsv; cd ..
 ./make.sh
 
 export CURATED='math_349'
+cd data; rm -f curated.tsv; ln -s VSAC/$CURATED.tsv curated.tsv; cd ..
 ./make.sh
 
 #export CURATED='medrt'
 #./make.sh
 
 export CURATED='mitre'
+cd data; rm -f curated.tsv; ln -s VSAC/$CURATED.tsv curated.tsv; cd ..
 ./make.sh
 
 export CURATED='CancerLinQ_non'
+cd data; rm -f curated.tsv; ln -s VSAC/$CURATED.tsv curated.tsv; cd ..
 ./make.sh
 
 export CURATED='mdpartners_non'
+cd data; rm -f curated.tsv; ln -s VSAC/$CURATED.tsv curated.tsv; cd ..
 ./make.sh
 
 export CURATED='atc_non'
+cd data; rm -f curated.tsv; ln -s VSAC/$CURATED.tsv curated.tsv; cd ..
 ./make.sh
 
 #export CURATED='medrt_non'
 #./make.sh
 
 export CURATED='cliniwiz_keywords'
+cd data; rm -f curated.tsv; ln -s VSAC/$CURATED.tsv curated.tsv; cd ..
 ./make.sh
 
 #export CURATED='keywords'
-#./make.sh
+#./make.sh

cumulus_library_opioid/vocab/vsac.py

@@ -1,3 +1,4 @@
+import argparse
 import json
 import pathlib
 import sys
@@ -13,7 +14,7 @@
     "CancerLinQ_non": "2.16.840.1.113762.1.4.1260.173",
     "CancerLinQ": "2.16.840.1.113762.1.4.1116.449",
     "cliniwiz_keywords": "2.16.840.1.113762.1.4.1200.163",
-    "clinwiz": "2.16.840.1.113762.1.4.1200.14",
+    "cliniwiz": "2.16.840.1.113762.1.4.1200.14",
     "ecri": "1.3.6.1.4.1.6997.4.1.2.234.999.3.2",
     "impaq": "2.16.840.1.113762.1.4.1196.87",
     "lantana": "2.16.840.1.113762.1.4.1046.241",
@@ -30,7 +31,9 @@
 def download_oid_data(
     source_vocab: str,
     *,
-    config: base_utils.StudyConfig,
+    api_key: str | None = None,
+    config: base_utils.StudyConfig | None = None,
+    force_upload:str | None = None,
     path: pathlib.Path | None = None,
 ) -> bool:
     """Fetches code definitions (assumed to be RXNorm coded) from VSAC
@@ -40,10 +43,13 @@ def download_oid_data(
     :keyword path: A path to write data to (default: ../data)
     :returns: True if file created, false otherwise (mostly for testing)
     """
+    if config:
+        api_key=config.umls_key
+        force_upload=config.force_upload
     if not path:
         path = pathlib.Path(__file__).parent.parent / "data"
     path.mkdir(exist_ok=True, parents=True)
-    if not config.force_upload and (path / f"{source_vocab}.parquet").exists():
+    if not (force_upload) and (path / f"{source_vocab}.parquet").exists():
         print(f"{source_vocab} data present, skipping download")
         return False
     if source_vocab not in VSAC_OIDS.keys():
@@ -52,7 +58,7 @@ def download_oid_data(
         # TODO: Impove this exist message when porting to library
         sys.exit(f"No available OID for {source_vocab}.")
     print(f"Downloading {source_vocab} to {path}")
-    api = umls.UmlsApi(api_key=config.umls_key)
+    api = umls.UmlsApi(api_key=api_key or api_key)
     output = []
 
     response = api.get_vsac_valuesets(oid=VSAC_OIDS[source_vocab])
@@ -92,3 +98,28 @@ def get_vsac_stewards(config:base_utils.StudyConfig) -> list[str]:
             "'--option steward:name' to specify which value sets to use.\n\n"
             f"{VALID_MSG}"
         )
+
+def main(cli_args=None):
+    """Temporary CLI interface"""
+    parser = argparse.ArgumentParser()
+    parser.add_argument(
+        "source_vocab", help="Name of vocab to look up codes for", default=None
+    )
+    parser.add_argument("--api_key", help="UMLS api key", default=None)
+    parser.add_argument(
+        "--force-upload",
+        help="Force redownloading of data even if it already exists",
+        action="store_true",
+    )
+    parser.add_argument("--path", help="optional path to write data to", default=None)
+    args = parser.parse_args(cli_args)
+    return download_oid_data(
+        args.source_vocab,
+        api_key=args.api_key,
+        force_upload=args.force_upload,
+        path=pathlib.Path(args.path),
+    )
+
+
+if __name__ == "__main__":
+    main()