README_Maintenance.md

Maintenance

This README contains useful information for maintaining the ATLAS system after it has been integrated.

Differential Donor Import

After the initial full load of donors into Atlas, the donor store can be kept in-sync with the external donor register via differential donor updates. These import files have the same schema as full files, but the updateMode must be set to diff.

Further reading:

• Donor Import README for more info on donor import - both, full and diff.

Donor Checkers

Functions are available to check whether the donor info held in the Atlas donor store is as expected. The consumer can then generate donor updates (i.e., diff donor import files) where differences have been reported.

Further reading:

• Donor Import README for info on how to use the donor checker functions.

Data Refresh Job

This job updates the HLA Metadata Dictionary (HMD) to the latest released IMGT/HLA nomenclature version, and re-processes donor HLA typings within the matching algorithm to this new HLA version. If configured to run automatically, it will run every time a new IMGT/HLA release is detected (usually, quarterly); else, it can be configured to run manually.

Further reading:

• Integration README for information on how to run the job.
• Support README for how to manually trigger the job and what to do if it fails.
• Configuration README for various settings that control how the job is run and resource allocation.
• Matching Algorithm README for info about the data processing itself.

HLA Haplotype Frequency Set (HFS) Import

• After the initial load of HFS files, new files can be imported at any time to update the data held by Atlas. File schema is identical to that of initial upload.
• If the combination of donPool and ethn within the new file already exists in the Atlas HFS store, then the previous set will be made inactive, and the new set will be used in match prediction for that population
• To update the "global" HFS, then upload a new file with donPool and ethn each set to null or [] (empty array).
• As there is no hard delete in the upload process, no sets will ever be deleted during ongoing operation. It may be desirable to manually delete older sets, to free up database space and keep database operations quick.

Further reading:

• Integration README for information on how to import HFS files.
• Match Prediction README for how HFS are selected during match prediction.
• Support README for how to handle failures during HFS import and rollback advice.