Investigate OMOP to CSV solution

[frankcorneliusmartin]

Currently the algorithms execute the query themselves (using the cohort id from previous created cohorts). Alternatively we could let the cohort create function overwrite the a default.csv. Then all other algorithms can stay vanilla.

Each time you want to use a different cohort, you need to run a specific job to overwrite the CSV. (or we can use whatever CSV as we can just read the folders, but that complicates stuff in the UI i think). Maybe we can also use several slots in the node config, that way we always have the option to store the cohort in a different slot (allowing to compare cohorts as well). Downside is that the user needs to do the book keeping.

Column names are reported on startup. Since the column names (variables) do not change very often we could create a default csv containing the headers. In case of an update we would have to put a new CSV in the nodes and reboot. Alternatively we could register it as a custom CSV and create a retrieval task for it.. Or just not use the official UI.

• [x] Check in the code that the algorithm has write permissions on the csv files
• [x] Check how we can make a custom variable retrieval job
• [x] Check that we can make several databases

[frankcorneliusmartin]

Currently a csv is exported into the /mnt/data folder. We are using a custom UI and job to manage these files. The only downside of this problem is that we need to modify the algorithms in order to find the correct cohort. Thus we do need a minor modification on the algorithms in order to make them work for this registry. I should be a very simple modification though.