added MRSTY

OHDSI · Aug 10, 2020 · 44d14ca · 44d14ca
1 parent 135b485
commit 44d14ca
Show file tree

Hide file tree

Showing 4 changed files with 22 additions and 4 deletions.
diff --git a/UMLS/create_source_tables.sql b/UMLS/create_source_tables.sql
@@ -147,6 +147,17 @@ CREATE TABLE SOURCES.MRREL
   FILLER_COLUMN INT
 );
 
+DROP TABLE IF EXISTS SOURCES.MRSTY;
+CREATE TABLE SOURCES.MRSTY
+(
+  CUI       CHAR(8),
+  TUI       VARCHAR(10),
+  STN       VARCHAR(100),
+  STY       VARCHAR(1000),
+  ATUI      VARCHAR(11),
+  CVF       VARCHAR(50)
+);
+
 CREATE INDEX X_MRSAT_CUI ON SOURCES.MRSAT (CUI);
 CREATE INDEX X_MRCONSO_CODE ON SOURCES.MRCONSO (CODE);
 CREATE INDEX X_MRCONSO_CUI ON SOURCES.MRCONSO (CUI);
@@ -158,4 +169,5 @@ CREATE INDEX X_MRCONSO_SDUI ON SOURCES.MRCONSO (SDUI);
 CREATE INDEX X_MRCONSO_STR ON SOURCES.MRCONSO (STR);
 CREATE INDEX X_MRCONSO_SUI ON SOURCES.MRCONSO (SUI);
 CREATE INDEX X_MRREL_AUI ON SOURCES.MRREL (AUI1, AUI2);
+CREATE INDEX X_MRSTY_CUI ON SOURCES.MRSTY (CUI);
 ALTER TABLE SOURCES.MRCONSO ADD CONSTRAINT X_MRCONSO_PK PRIMARY KEY USING INDEX X_MRCONSO_PK;
diff --git a/UMLS/readme.md b/UMLS/readme.md
@@ -5,13 +5,14 @@ Prerequisites:
 - Working directory UMLS.
 
 1. Run create_source_tables.sql
-2. Download UMLS Metathesaurus Files (for example umls-YYYYAB-metathesaurus.zip) from http://www.nlm.nih.gov/research/umls/licensedcontent/umlsknowledgesources.html
-3. Unpack files 
+2. Download Full UMLS Release Files (for example umls-YYYYAA-full.zip) from http://www.nlm.nih.gov/research/umls/licensedcontent/umlsknowledgesources.html
+3. Unpack files
 MRCONSO.RRF
 MRHIER.RRF
 MRMAP.RRF
 MRSMAP.RRF
 MRSAT.RRF
 MRREL.RRF
+MRSTY.RRF
 
 4. Run in devv5 (with fresh vocabulary date): SELECT sources.load_input_tables('UMLS',TO_DATE('20180507','YYYYMMDD'),'2018AA');
diff --git a/working/packages/load_input_tables/load_input_tables.sql b/working/packages/load_input_tables/load_input_tables.sql
@@ -13,7 +13,7 @@ begin
   pVocabularyPath=pVocabularyPath||pVocabularyID||'/';
   case pVocabularyID
   when 'UMLS' THEN
-      truncate table sources.mrconso, sources.mrhier, sources.mrmap, sources.mrsmap, sources.mrsat, sources.mrrel;
+      truncate table sources.mrconso, sources.mrhier, sources.mrmap, sources.mrsmap, sources.mrsat, sources.mrrel, sources.mrsty;
       alter table sources.mrconso drop constraint x_mrconso_pk;
       drop index sources.x_mrsat_cui;
       drop index sources.x_mrconso_code;
@@ -25,6 +25,7 @@ begin
       drop index sources.x_mrconso_str;
       drop index sources.x_mrconso_sui;
       drop index sources.x_mrrel_aui;
+      drop index sources.x_mrsty_cui;
       /*
       UMLS can contain characters like single quotes and double quotes, but PG uses them as a service characters
       So we specifying a quote character that should never be in the text: E'\b' (backspace)
@@ -35,6 +36,7 @@ begin
       execute 'COPY sources.mrsmap (mapsetcui,mapsetsab,mapid,mapsid,fromexpr,fromtype,rel,rela,toexpr,totype,cvf,vocabulary_date) FROM '''||pVocabularyPath||'MRSMAP.RRF'' delimiter ''|'' csv quote E''\b''';
       execute 'COPY sources.mrsat FROM '''||pVocabularyPath||'MRSAT.RRF'' delimiter ''|'' csv quote E''\b''';
       execute 'COPY sources.mrrel FROM '''||pVocabularyPath||'MRREL.RRF'' delimiter ''|'' csv quote E''\b''';
+      execute 'COPY sources.mrsty FROM '''||pVocabularyPath||'MRSTY.RRF'' delimiter ''|'' csv quote E''\b''';
       update sources.mrsmap set vocabulary_date=COALESCE(pVocabularyDate,current_date), vocabulary_version=COALESCE(pVocabularyVersion,pVocabularyID||' '||current_date);
 
       CREATE INDEX x_mrsat_cui ON sources.mrsat (cui);
@@ -48,13 +50,15 @@ begin
       CREATE INDEX x_mrconso_str ON sources.mrconso (str);
       CREATE INDEX x_mrconso_sui ON sources.mrconso (sui);
       CREATE INDEX x_mrrel_aui ON sources.mrrel (aui1, aui2);
+      CREATE INDEX x_mrsty_cui ON sources.mrsty (cui);
       ALTER TABLE sources.mrconso ADD CONSTRAINT x_mrconso_pk PRIMARY KEY USING INDEX x_mrconso_pk;
       analyze sources.mrconso;
       analyze sources.mrhier;
       analyze sources.mrmap;
       analyze sources.mrsmap;
       analyze sources.mrsat;
       analyze sources.mrrel;
+      analyze sources.mrsty;
   when 'CIEL' then
       set local datestyle='ISO, DMY'; --set proper date format
       truncate table sources.concept_ciel, sources.concept_class_ciel, sources.concept_name, sources.concept_reference_map, sources.concept_reference_term, sources.concept_reference_source;

diff --git a/working/packages/vocabulary_download/bash_functions_umls.sql b/working/packages/vocabulary_download/bash_functions_umls.sql
@@ -19,7 +19,8 @@ BEGIN
     mv "MRMAP.RRF" "$1" && \
     mv "MRSMAP.RRF" "$1" && \
     mv "MRSAT.RRF" "$1" && \
-    mv "MRREL.RRF" "$1"
+    mv "MRREL.RRF" "$1" && \
+    mv "MRSTY.RRF" "$1"
     $BODY$
     LANGUAGE 'plsh'
     SECURITY DEFINER;