Checking & refreshing metadata changes

In some occasions, the analytics team may need to take specific actions to check or update changes in the datamarts. Examples in which this may happen may include the following:

• Consent withdrawal
  • Participants may voluntarily request to withdraw from research studies or other projects. Regulations indicate that existing metadata must be updated to reflect the withdrawal, which requires following the correct process to update our datamarts.
• Fingerprinting and gender related changes
  • Requests for fingerprinting reanalyses and gender related changes may occur for several reasons such as missing FP assays, sample swaps, and collaborator requests. These changes may not refresh automatically in the datamarts for which refresh pushes may be necessary. 

The main steps a user may need to follow in these occasions include checking metadata changes and pushing datamart refreshes. The processes explained below are meant to help the user take step-by-step actions if and when required. A user may not need to take the steps in all processes.

Table of Contents

1. Checking Metadata Changes

Download the CSV or XLS file from the SUPPORT ticket (they should be interchangeable). Copy the distinct "Old" Collaborator Sample IDs to paste them into the SQL queries below.

1.1. Check whether PDO_STAR is updated

1. Log in to REPORTING@SEQPROD on SQL Navigator and run the following query:

   SELECT DISTINCT s.bsp_collaborator_sample_id, 
   				s.collaborator_participant_id, 
   				s.aggregation_project, 
   				s.last_aggregation_date, 
   				s.external_sample_id, 
   				s.pdo_name, 
   				s.pdo_sample_id, 
   				s.sample_delivery_status, 
   				s.sample_is_billed
   FROM cognos.pdo_star5 s 
   WHERE s.bsp_collaborator_sample_id IN ('oldcollabSMID1','oldcollabSMID2')
   
   

2. If no records are returned, then no further action is needed (datamarts have already been refreshed)
3. If records are returned, then either there are supposed to be some samples remaining (this is less frequent) or datamarts need refreshing.
4. Before refreshing PDO STAR, first check to see if SLXRE_READGROUP_METADATA is updated (section 1.2 below).
5. If readgroup metadata is updated, then the PDO STAR datamart needs a refresh (see sections 2.1-2.3 below). Note the PDO_NAME of any records returned.

1.2. Check whether SLXRE_READGROUP_METADATA is updated

1. Log in to REPORTING@SEQPROD on SQL Navigator and run the following query:

   SELECT DISTINCT a.collaborator_sample_id,
   				a.sample_id bsp_sample_id,
   				a.project, a.research_project_id,
   				a.product_order_key,
   				a.product_order_sample,
   				a.collaborator_participant_id,
   				a.flowcell_barcode,
   				a.run_name
   FROM cognos.slxre_readgroup_metadata a
   WHERE a.collaborator_sample_id  IN ('oldcollabSMID1','oldcollabSMID2')

2. If no records are returned, then no further action is needed (datamarts have already been refreshed)

3. If records are returned, then the datamart needs a refresh (section 2.4 below). Note the FLOWCELL_BARCODE of any records returned.

 1.3. Check whether there's updated data to be added

1.  Check for metrics in DB with the following queries:

   SELECT *
   FROM analytics.bsp_sample
   WHERE  collaborator_sample_id IN 'oldcollabSMID1, oldcollabSMID2'
   ;
   SELECT *
   FROM metrics.basecalling_analysis ba 
   WHERE ba.flowcell_barcode IN ('HTH3KBBXX')
   AND ba.metrics_type IN ('Indexed', 'Unindexed')
   ;
   SELECT *
   FROM metrics.picard_analysis ba 
   WHERE ba.flowcell_barcode IN ('HTH3KBBXX')
   AND ba.metrics_type IN ('Indexed', 'Unindexed')
   ;
   --Understand when flowcells were updated in Analytics DM
   SELECT *
   FROM cognos.slxre_rghqs_targets t 
   WHERE t.flowcell_barcode = 'HTH3KBBXX'
   ;

1.4. Check aggregations

1.  Check for aggregations in DB with the following queries:

SELECT *
FROM cognos.slxre2_pagg_sample a 
WHERE a.SAMPLE IN ('UNKNOWN_SubmittedAs_33433.wb','UNKNOWN_SubmittedAs_33514.wb')
;
SELECT a.sample_ids plated_samples, a.individual_name, a.*
FROM cognos.slxre2_pagg_sample a 
WHERE a.SAMPLE IN ('33433.wb','33514.wb')
;
-- check whether the same read group is aggregated under old and new collab sample id 
SELECT is_latest, rg.flowcell_barcode, rg.lane, rg.molecular_barcode_name, count(*),
concat_string_csv(DISTINCT a.SAMPLE )
FROM metrics.aggregation a,
metrics.aggregation_read_group rg  
WHERE a.SAMPLE IN ('UNKNOWN_SubmittedAs_33433.wb','33433.wb')
 AND a.is_latest <> 0
 AND a.id = rg.aggregation_id
GROUP BY rg.flowcell_barcode, rg.lane, rg.molecular_barcode_name, is_latest

NOTE: is_latest should be 2 for anything with consent withdrawn (2 signifies BAM is deleted)

1.5. Check whether Sample datamarts have been refreshed

1. If refreshes aren't showing updated data, it may be because the sample ETL is down. If LIMS team has done their part, but the bsp.analytics_sample and analytics.bsp_sample DMs are not updated, check the ETL agent here: http://analytics:8090/etl_runs?status=Succeeded&status=Failed&status=Running&status=Abandoned&agent=analytics.tiger.agents.BspSample&action=Refresh

1.6. Resolving DataChecker Errors - PK violations of Aggregation DMs

1. In cognos, check whether PK violators have already been picked up by a subsequent ETL
   

   SELECT e.error_timestamp, la.TIMESTAMP etl_timestamp, e.*
   FROM err$_slxre2_pagg_library e
   LEFT JOIN slxre2_pagg_library la ON
   la.project = e.project
   AND la.SAMPLE = e.SAMPLE
   AND la.LIBRARY = e.LIBRARY
   AND la.data_type = e.data_type
   WHERE e.resolution_timestamp IS NULL

2. If etl_timestamp is after the last error_timestamp , the error record can be resolved(use the query below). If etl_timestamp is missing or is before error_timestamp, then follow up with Pipeline team, send them specific (project, sample, librarym data type)
   

   UPDATE err$_slxre2_pagg_library
   SET resolution_timestamp = SYSDATE
   WHERE resolution_timestamp IS NULL

3. If a subset of violators were picked up by the ETL, adjust the WHERE clause in the UPDATE statement to just those entries

2. PDO STAR, RGHQS, and Read Group Metadata Datamart Refresh Push

2.1. PDOSTAR refresh Option 1 - Using RunEtl tool (Also RGHQS refresh)

1. Instructions here

2.2. PDOSTAR refresh Option 2 - UNIX, using a query (using "Old" Collaborator Sample IDs)

1. Log in to your favorite SSH tool (e.g. SecureCRT)
   
   1. connect to "analytics" server
   2. enter the following: /home/unix/analytics/TigerETL3/runEtlShell.sh

   3. paste & run the following etl-command once, pasting the "Old" Collaborator Sample IDs in the query below

      import analytics.tiger.agents.PdoStar._
      AnalyticsEtlDB("analytics.tiger.agents.PdoStar.fromSQL", collSamplesToPdoSamples(List("oldcollabSMID1","oldcollabSMID2")) flatMap DataSetEtl() flatMap pipeline)

      NOTE: These are supposed to be two lines (the first ends with an underscore). Once you hit enter, the command takes a little while. You'll know it's complete when you get the scala prompt again.

2.3. PDOSTAR refresh Option 3 - UNIX, refresh using a query (using PDO_NAMEs)

1. Log in to your favorite SSH tool (e.g. SecureCRT)
   
   1. connect to "analytics" server (command: ssh analytics)
   2. enter the following: /home/unix/analytics/TigerETL3/runEtlShell.sh

   3. paste & run the etl-command separately for each PDO, changing PDO-12480 to the relevant PDO returned above.

      AnalyticsEtlDB("analytics.tiger.agents.PdoStar.fromSQL", deltaFromSql("""SELECT DISTINCT a.pdo_name||','||a.pdo_sample_id FROM cognos.pdo_star5 a WHERE a.pdo_name = 'PDO-12480'""") flatMap DataSetEtl() flatMap analytics.tiger.agents.PdoStar.pipeline)

   4. The command takes a little while. You'll know it's complete when you get the scala prompt again

2.4. PDO STAR review to ensure it did refresh

1. Rerun the PDO_STAR query above. This refresh is immediate, so you can run the query right away.
2. If it returns nothing, the DM is all set.
3. If it still returns records, ensure that that is unexpected behavior and try the refresh again. (occasionally, there are other samples that need to remain with the "old" IDs, as is usually the case in sample swaps)
   1. If it still returns unexpected records, talk with Nasko.

2.5. ReadgroupMetadata DM Refresh in SEQPROD Database

1. Log in to COGNOS@SEQPROD on SQL Navigator

2. Request a refresh for RunMercuryAgent by issuing the following query:
   

   BEGIN
   cognos.data_warehouse_api.request_refresh(P_FLOWCELL_BARCODE=>'XYZ', P_REQUESTED_BY=>'JohnDoe', P_COMMENTS=>'Test', P_DEFERRAL_ANALYSIS_DATE=>null);
   END;

   Note: No COMMIT command needed after this particular call - it's implicitly done for you. Just replace the XYZ with a flowcell barcode. Leave all other parameters there and unchanged.

   If you have multiple flowcells, you may run the query as follows, adding additional lines after BEGIN:

   BEGIN
   cognos.data_warehouse_api.request_refresh(P_FLOWCELL_BARCODE=>'XYZ1', P_REQUESTED_BY=>'JohnDoe', P_COMMENTS=>'Test', P_DEFERRAL_ANALYSIS_DATE=>null);
   cognos.data_warehouse_api.request_refresh(P_FLOWCELL_BARCODE=>'XYZ2', P_REQUESTED_BY=>'JohnDoe', P_COMMENTS=>'Test', P_DEFERRAL_ANALYSIS_DATE=>null);
   cognos.data_warehouse_api.request_refresh(P_FLOWCELL_BARCODE=>'XYZ3', P_REQUESTED_BY=>'JohnDoe', P_COMMENTS=>'Test', P_DEFERRAL_ANALYSIS_DATE=>null);
   ...
   cognos.data_warehouse_api.request_refresh(P_FLOWCELL_BARCODE=>'XYZX', P_REQUESTED_BY=>'JohnDoe', P_COMMENTS=>'Test', P_DEFERRAL_ANALYSIS_DATE=>null);
   END;
   
   

2.5. Wait for RunMercury agent and recheck.

The RunMercury agent is run on a schedule. After the next run, rerun the SELECT SQL clause in step 1.2 to ensure that the records have been refreshed and are not returned with the old collab sample id.

http://analytics:8090/etl_runs?status=Succeeded&status=Failed&status=Running&status=Abandoned&status=Halted&agent=analytics.tiger.agents.RunMercury&action=Refresh

3. Related pages

Performing metadata changes

Requesting sample metadata changes

Tiger3 ETL Shell Commands (refreshes, etc.)