AMIS Integration

[thk123]

Implementing the wrapper layer between the model and AMIS. This is inside a folder called amis_integration inside sch_simulation. It contains a README explaining how to use

The Model Wrapper

Duplicates a lot of the logic of Example_Run.py, but exposes all the customisable points so that the R script dictates the run.

The entry point is run_model_with_parameters which is expected to be called from R with appropriate parameters. It runs on a single thread - this might be something that we want to change.

The R code

There is a script sth_fitting.R which contains the bespoke parameters for a specific fit (one could imagine having another script called sch_fitting.R with different parameters. It uses amis_integration.R which handles the python/R interface, as well as making sure it uses the right Python virtual env.

At the moment it contains totally dummy prior, prevalence maps etc - this will be done by Evandro.

Fixes NTD-Modelling-Consortium/endgame-project#3

To do:

• [x] Get the virtual environment programmatically in the R script
• [x] Allow the R script to drive what years we are trying to fit to
• [x] Fix typo in name of the R or Python script (whichever is wrong!)
• [x] Reference specific AMIS version
• [x] Add a trivial R test to check the utility
• [x] Fix python version (3.8 does not provide the @cache functionality
• [x] Fix the import
• [x] Test on 3.10 rather than 3.12
• [x] Ensure cache is working

[thk123]

@mattg3004 I've put this back as a draft as I've realised there are a couple of tidy up things I should do that were caught by Evandro trying to use this, so I'll do that and then ping you when it is finally ready for review.

[thk123]

Raised NTD-Modelling-Consortium/endgame-project#14 for the multi point fitting

[thk123]

For now assuming will be run with Rscript in the correct working directory, have raised NTD-Modelling-Consortium/endgame-project#15 for making the fitting work without precise working directory stuff.

[thk123]

@mattg3004 I am happy for this to be merged, if you are happy with (ie understand it, can't see any problems), then could you "approve" it, by going to the files changed tab and submit an approving review: Thanks 💚

[thk123]

@mattg3004 Thanks for reviewing - I've asked Raiha to review the R side of things.