The kids23-team1's discuss from stjude-biohackathon

I'll set this up.

This should have ~8 cores and 32G RAM, and should have a 20G disk mounted as a filesystem (this will act like our tartan).

The schema should contain the following tables that represent a minimal design for what we are trying to accomplish:

Order Table that contains 1 || 2 samples and an analysis suite - FK on sample id + analysis suite id
Analysis Suites Table that has suite names corresponding to a set of analyses - 1 to many with Analysis ID
Analysis table that contains all possible analyses (only a select few will correspond to a given analysis suite)
Analysis run table - FK on analysis id + sample id
Sample table

This is all that comes to mind top of head. The flow of the process is:

A lab member places an order for analysis on a sample / sample pair
the program looks up the analysis suite for that order type
If neither sample has any previous analysis, skip 4-7
if there are 2 samples, it checks to see if either of the samples already has analyses run on it
if a sample does, it checks to see which tier of filesystem those analysis runs are in
if they are in hot tier, they are pulled back
if they are in archive tier, a rehydration is initiated and tracked until complete. when complete, it is moved back to hot tier
when a single analysis completes, the program should copy it to hot tier (but not delete) and update the database
when all analyses complete, we want to update the database and delete the data from local filesystem

It is entirely possible I am forgetting things, this should serve as a nice template though

Everyone on the team should be able to view the resource group (and resources) in the portal, as well as create them.

This ticket will specifically be for deploying the resource, the functionality for the azure function will be in another ticket.

stjude-biohackathon / kids23-team1 Goto Github PK