more work on linking market data

RxMix/ASSOICATING NCTIDs to NDCs and Marketing dates.sql

@@ -1,7 +1,3 @@
-/*Get relationships between brands and branded drugs and packs*/
-select * from rxnorm_migrated.rxnorm_relations rr 
-where tty1 = 'BN' and tty2 in ('SBD', 'BPCK')
-limit 100;
 
 /*get all the ndc codes associated with an rxcui
  * Same as query
@@ -11,17 +7,85 @@ limit 100;
  * Based on http://will-office:4000/RxNav/search?searchBy=RXCUI&searchTerm=1668240
  * it appears that this rxcui is a sbd or bpck (branded drug or pack)
  * 
- * If I grab every brand, then every branded drug or pack associated with that drug and then every 
- * */
-select * from ALLNDC_HISTORY ah 
-where RXCUI ='1668240'
-and SAB='RXNORM'
-;
-/** 
+
  * If I grab every brand, then every branded drug or pack associated with that drug and then attach that to the nsde data I would get the marketing dates required.
- * trial -> mesh_term -> IN/MIN (rxcui) -> BN (rxcui) -> SBD/BPCK (rxcui) -> ndc11 -> nsde (marketing dates)
+ * --get brand names
+ * trial -> mesh_term -> IN/MIN (rxcui) -> BN (rxcui) 
+ * -- associate brand names to marketing dates
+ * BN (rxcui) --> SBD/BPCK (RXCUI) --> ndc11 --> nsde 
  * */
 
 /*
  * I do need to figure out a way to change the date types when importing into postgres. In mariadb they ar mmYYYY wheras in the jsonapi they are YYYYmm but I want is YYYY-mm-01
- */*/
+ */
+
+---assoicate ingredients, brands, and approved packaging RXCUIs.
+create temp table trial_to_pack_rxcui as
+with trialncts as (
+        SELECT DISTINCT nct_id  FROM history.trial_snapshots TS
+    )
+SELECT 
+    bi.nct_id ,
+    bi.downcase_mesh_term, 
+    rp.rxcui  AS drug_rxcui,
+    rn.tty1,
+    rn.rxcui1 as ingredient_rxcui, --ingredients
+    rn.tty2 ,
+    rn.rxcui2 as brand_or_pack_rxcui --brand or pack
+FROM ctgov.browse_interventions bi
+    LEFT OUTER JOIN rxnorm_migrated.rxnorm_props AS rp
+        on bi.downcase_mesh_term  = rp.propvalue1  --Link drug ingredient
+    LEFT OUTER JOIN rxnorm_migrated.rxnorm_relations AS rn
+        on rp.rxcui = rn.rxcui1 --Grab brand names
+WHERE
+    bi.nct_id in (
+        SELECT nct_id  FROM trialncts
+        )
+    and
+    bi.mesh_type='mesh-list'
+    and
+    rn.tty1 in ('IN','MIN')
+    and 
+    rn.tty2 = 'BN'
+;
+
+
+--link brand names to drug applications (NDA/ANDA etc)
+select rr.rxcui1 as BN, rr.rxcui2 as pack, ah.ndc as pack_ndc11
+from 
+    rxnorm_migrated.rxnorm_relations rr 
+    left outer join rxnorm_migrated."ALLNDC_HISTORY" as ah 
+        on rr.rxcui2 = ah.rxcui 
+where 
+    tty1 = 'BN' 
+    and
+    tty2 in ('SBD', 'BPCK')
+    and 
+    ah.sab='RXNORM'
+    
+;
+
+
+---associate NDAs/ANDAs to marketing start dates
+---Get start of coverage periods for NSDE dates grouped by arbitrary grouping.
+SELECT n.application_number_or_citation, count(*), min( marketing_start_date )
+FROM spl.nsde as n
+where product_type = 'HUMAN PRESCRIPTION DRUG'
+group by n.application_number_or_citation ;
+
+
+---For a given date, find which NDAs/ANDAs were active were active.
+SELECT n.application_number_or_citation, count(*)
+FROM spl.nsde as n
+where 
+    product_type = 'HUMAN PRESCRIPTION DRUG'
+    and
+    marketing_start_date < '2010-05-01'
+    and
+    marketing_end_date > '2010-05-01'
+group by n.application_number_or_citation ;
+
+
+
+
+

RxMix/migrate_mysql2pgsql.py

@@ -133,9 +133,6 @@ if __name__ == "__main__":
                 #insert the data with page_size
                 extras.execute_values(pcurse,psql_insert,argslist=a,template=template, page_size=1000)
                 """ 
-                ISSUE HERE ^^^^^ somehow execute values isn't separating over dictionaries very well
-                https://www.psycopg.org/docs/extras.html#psycopg2.extras.execute_batch
-                maybe replace with execute_batch?
                 """