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 their participation from specific ongoing research studies or other projects. Regulations indicate that any existing metadata must be updated to reflect such withdrawal request. These changes may not be readily available until the right steps are followed to update the metadata on the 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 includes 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.
...
- 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
- 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$_slxre2_pagg_library e
LEFT JOIN slxre2_pagg_library la ON
la.project = e.project
AND la.SAMPLE = e.SAMPLE
AND la.LIBRARY = e.LIBRARY
AND la.data_type = e.data_type
WHERE e.resolution_timestamp IS NULL - If etl_timestamp is after the last error_timestamp , the error record can be resolved(use the query below). If etl_timestamp is missing or is before error_timestamp, then follow up with Pipeline team, send them specific (project, sample, librarym data type)
UPDATE err$_slxre2_pagg_library
SET resolution_timestamp = SYSDATE
WHERE resolution_timestamp IS NULL - 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
...
- Rerun the PDO_STAR query above. This refresh is immediate, so you can run the query right away.
- If it returns nothing, the DM is all set.
- 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)
- If it still returns unexpected records, talk with Nasko.
2.5.
...
ReadgroupMetadata DM Refresh in SEQPROD Database
Tip | ||||
---|---|---|---|---|
| ||||
Review Analytics ETL Framework for more details on this section |
...