Atlas Notes

General Notes

• The psql database must be created and the webapi_admin must be given full control over the database.
• The application is then configured with the connection string for the database, and applies database migrations the first time it starts and can connect to the database.
• Once the migrations are applied, data source configuration needs to be done.
• Additional security configuration may need to be done depending on the security configuration and the deployment process.

Configuring Sources

Once the database migrations have been applied, records will need to be inserted to tell the WebAPI where to find OMOP/CDM data.

You can find a script that does this in the atlas terraform repo: https://dev.azure.com/wustl-i2/Platform Engineering/_git/atlas-terraform?path=/scripts/configure_app.sql&version=GBmain&_a=contents
This script waits to ensure that the database migrations are completed before inserting the configuration for the OMOP Demo data.

Ideally we will have (at least) two existing OMOP data sources in place before spinning up an instance of WebAPI. We should configure each (production) instance to point to both OMOP demo data and OMOP production data.

To enable testing, we will load a demo OMOP data set onto a postgres server to make it easier to connect WebAPI to the data. Eventually we will move the demo data to Databricks and point WebAPI to Databricks for both the demo and production OMOP data.

Querying data from Databricks requires some additional setup of the WebAPI container to tell it to use the Databricks JDBC driver when querying the CDM. Work pending...

Loading CDM

This document describes how to use the OHDSI tools to generate a sample OMOP data set: https://github.com/microsoft/OHDSIonAzure/blob/main/docs/PrepareSyntheticOMOP.md

This should only need to be done one time and placed onto a postgres server. Once we have the sample data, we can point WebAPI to the database and copy it to Databricks as well.

Ulterior Goals

1. get an example OMOP dataset
2. stick it somewhere inside of postgres
3. stick it somewhere inside of databricks

Technical Extras

• Existing Postgres Dev Database see:
  • https://portal.azure.com/#@gowustl.onmicrosoft.com/asset/Microsoft_Azure_KeyVault/Secret/
  • https://i2-atlas-dev-kv.vault.azure.net/secrets/webapi-jdbc-conn-str/c82d99224e3a4e9ebac6216d02c9dd53
• https://github.com/synthetichealth/synthea/wiki
• Azure Resource Group: i2-atlas-dev-rg

See Also

WebAPI Repo
Broadsea Repo
Repo with sample data example
Atlas App - Dev Instance