Drug Domain‎ > ‎

Multum

Vocabulary_ID=16

The following describes the procedure to update the Multum vocabulary (concepts, relationships, ancestry) with the latest release of the source.

The load process is a complete refresh of data. All new vocabulary entieties will be added. All vocabulary entities that do not exist in the latest files set will be marked as deprecated.

1.1 Scripts Transfer

All scripts and files listed at the bottom of this page should be transferred to specially created directory (eg /data/MULTUM).


 Script Name Description
16_create_schema.sql Create prestage-stage schema with tables and indexes
16_load_to_prestage.sh Load RxNorm row-data
16_transform_row_maps.sql  Convert and store in stage table SOURCE_TO_CONCEPT_MAP_STAGE
16_load_maps.sql  Load only new maps into SOURCE_TO_CONCEPT_MAP table, add invalid code information


1.2 Download data from source

The most current international release is located at the following URL: http://www.nlm.nih.gov/research/umls/rxnorm/docs/rxnormfiles.html

Historical archives located at: http://www.nlm.nih.gov/research/umls/rxnorm/docs/rxnormarchive.html

Note: Make sure you perform updates in chronological order. So, if you want to process a January and April release, execute the below steps for January, and after successful completion, for April.

Download the latest release file (e.g. RxNorm_full_07022012.zip)
Transfer downloaded zip file to
/data/MULTUM directory.

Zip file suffix contains release date (e.g. 07022012=02-Jul-2012)
Latest production zip file is ~136MB in size

Note: For the rest of the document we will be referring to current schema as MULTUM_20120702 you would have to adjust this entry to correspond to current release date file

1.3 Build Database Schema

All tables will be located in the schema MULTUM_<DateOfSourceFile> as following: MULTUM_20120131
The following steps will create a schema and will grant to a schema owner appropriate permissions. 


In addition to the schema, it will create tables to load raw data: RXNCONSO  as well as staging tables: 

CONCEPT_STAGE, CONCEPT_ANCESTOR_STAGE, 

CONCEPT_RELATIONSHIP_STAGE, 

SOURCE_TO_CONCEPT_MAP_STAGE

$ sqlplus System/<SystemPass>@DEV_VOCAB @16_create_schema.sql <MULTUM_Schema> <Pass_MULTUM_Schema>

Format of MULTUM_Schema should be MULTUM_YYYYMMDD (e.g. MULTUM_20120702)


Staging tables match Vocabulary schema version 4.0 with slight modification. Modifications made to the tables are:

CONCEPT_STAGE.CONCEPT_ID is nullable

          CONCEPT_RELATIONSHIP_STAGE

.REL_ID is nullable

1.4 Raw data load into Oracle tables

1.4.1 Extract files from the zip archive into the current directory

        $ unzip -u RxNorm_full_07022012.zip -d rxnorm_20120702


1.4.2 Verify that the following  files have been created:

      RXNCONSO.RRF    (Latest file size ~96MB)

   

        1.4.3    Load raw data into Oracle. 

        This sqlldr process will run for about 2min.

        $ 16_load_to_prestage.sh /data/MULTUM/rxnorm_20120702 <MULTUM_schema>/<Pass_MULTUM_Schema>


1.4.4 Copy data and intermediate files to backup area In this step, downloaded files will be archived to preserve source of the data. Discuss a location of backup area with your system administrator.         

        $ cp -Ru /data/MULTUM/*.zip /data/backup_area ; cp -Ru /data/MULTUM/*.log /data/backup_area

1.4.5 Verify that number of records loaded is equivalent to prior production load

                     $ sqlplus MULTUM_20120702/myPass@DEV_VOCAB

            Execute the following SQL commands:

    SELECT count(*) FROM RXNCONSO;
Current production row count is:     977,332 Rows


1.5 Loading Staging Tables from raw

        1.5.1 Convert and store in staging table maps

Loaded from raw staged data RXNCONSO into the staged data SOURCE_TO_CONCEPT_MAP_STAGE


 

$  sqlplus MULTUM_20120702/myPass@DEV_VOCAB @16_transform_row_maps.sql


1.5.2 Verify that number of records loaded is equivalent to prior production load

         In this step we will verify that number of active records in staging tables is valid prior to transfer data to historical DEV schema. 

    a. Number of Records in stage table

    b. Number of records in DEV schema not deleted (active, prior load production)

    c. How many records would be added to DEV schema table

            d. How many active DEV records will be marked for deletion


            In most cases  number of records in stage (a) should be greater than number of active records in production (b) 
            In rare occasion that might not be the case. It is responsibility of the developer running the scripts to verify that that is the the valid.



     $ sqlplus MULTUM_20120507/myPass@DEV_VOCAB



        Execute the following SQL commands:
SELECT '- Num Rec in stage' AS scr, COUNT (8) cnt
  FROM SOURCE_TO_CONCEPT_MAP_STAGE c
 WHERE c.SOURCE_VOCABULARY_ID IN (16)
UNION ALL
SELECT '- Num Rec in DEV not deleted' AS scr, COUNT (8) cnt
  FROM DEV.SOURCE_TO_CONCEPT_MAP d
 WHERE     d.SOURCE_VOCABULARY_ID IN (16)
       AND D.TARGET_VOCABULARY_ID IN (08)
       AND NVL (d.INVALID_REASON, 'X') <> 'D'
UNION ALL
SELECT '- How many records would be new in DEV added' AS scr, COUNT (8) cnt
  FROM SOURCE_TO_CONCEPT_MAP_STAGE c
 WHERE c.SOURCE_VOCABULARY_ID IN (16) AND c.TARGET_VOCABULARY_ID IN (08)
       AND NOT EXISTS
                  (SELECT 1
                     FROM DEV.SOURCE_TO_CONCEPT_MAP d
                    WHERE     d.SOURCE_VOCABULARY_ID IN (16)
                          AND c.SOURCE_CODE = D.SOURCE_CODE
                          AND d.SOURCE_VOCABULARY_ID = c.SOURCE_VOCABULARY_ID
                          AND d.MAPPING_TYPE = c.MAPPING_TYPE
                          AND d.TARGET_CONCEPT_ID = c.TARGET_CONCEPT_ID
                          AND d.TARGET_VOCABULARY_ID = c.TARGET_VOCABULARY_ID)
UNION ALL
SELECT '- How many DEV active will be marked for deletion' AS scr,
       COUNT (8) cnt
  FROM DEV.SOURCE_TO_CONCEPT_MAP d
 WHERE     d.SOURCE_VOCABULARY_ID IN (16)
       AND D.TARGET_VOCABULARY_ID IN (08)
       AND NVL (d.INVALID_REASON, 'X') <> 'D'
       AND d.VALID_START_DATE <
              TO_DATE (
                 SUBSTR (USER, REGEXP_INSTR (USER, '_[[:digit:]]') + 1, 256),
                 'YYYYMMDD')
       AND NOT EXISTS
                  (SELECT 1
                     FROM SOURCE_TO_CONCEPT_MAP_STAGE c
                    WHERE     c.SOURCE_VOCABULARY_ID IN (16)
                          AND c.SOURCE_CODE = D.SOURCE_CODE
                          AND d.SOURCE_VOCABULARY_ID = c.SOURCE_VOCABULARY_ID
                          AND d.MAPPING_TYPE = c.MAPPING_TYPE
                          AND d.TARGET_CONCEPT_ID = c.TARGET_CONCEPT_ID
                          AND d.TARGET_VOCABULARY_ID = c.TARGET_VOCABULARY_ID)
       AND EXISTS
              (SELECT 1
                 FROM SOURCE_TO_CONCEPT_MAP_STAGE c
                WHERE     d.SOURCE_CODE = c.SOURCE_CODE
                      AND d.SOURCE_VOCABULARY_ID = c.SOURCE_VOCABULARY_ID
                      AND d.MAPPING_TYPE = c.MAPPING_TYPE
                      AND d.TARGET_VOCABULARY_ID = c.TARGET_VOCABULARY_ID);

                Current Result:
- Num Rec in stage                                                       9558
- Num Rec in DEV not deleted                                        9691
- How many records would be new in DEV added               56
- How many DEV active will be marked for deletion            14


    1.5.3   Load new maps into DEV schema concept table,mark deprecated concepts as deleted. Valid start date, valid end date and invalid reason are left empty for now.
          Transfer records from SOURCE_TO_CONCEPT_MAP_STAGE to DEV.SOURCE_TO_CONCEPT_MAP table

    $  sqlplus MULTUM_20120702/myPass@DEV_VOCAB @16_load_maps.sql


ċ
16_create_schema.sql
(7k)
Serg Vereshagin,
Oct 28, 2013, 2:24 AM
ċ
16_load_maps.sql
(6k)
Serg Vereshagin,
Oct 28, 2013, 2:24 AM
ċ
16_load_to_prestage.sh
(5k)
Serg Vereshagin,
Oct 28, 2013, 2:24 AM
ċ
16_transform_row_maps.sql
(5k)
Serg Vereshagin,
Oct 28, 2013, 2:24 AM
Comments