MoDS Wrapper

SimPhoNy-based implementation of the MoDS wrapper.

The wrapper currently supports the Multi Objective Optimization feature (MOO). The wrapper workflow is as follows:

• semantic (CUDS) to syntactic (JSON) translation of the input data
• triggering the MOO simulation by sending the translated input data via HTTP request to the MoDS web agent
• checking the simulation status
• collection of the simulation outputs from the MoDS web agent
• syntactic (JSON) to semantic (CUDS) translation of the simulation results

The mock version of the MoDS web agent is additionally provided for testing and debugging purposes.

Requirements

• Python 3.8 (or higher)
• SimPhoNy

Installation - docker

A convenience docker_install.sh script has been provided that creates the cmcl/sim_cmcl_mods_wrapper image.

    # build the cmcl/sim_cmcl_mods_wrapper image
    docker_install.sh

Installation - local

Please follow these steps to install the wrapper on your machine (using the virtual environment is highly recommended):

    # install the mods wrapper
    python -m pip install .
    # install mods simple ontology
    pico install ontology.mods.yml
    # (optional) install the wrapper tests requirements
    python -m pip install -r tests/test_requirements.txt

Usage

The mods wrapper can be used by either running it in a docker container or locally. In case of the docker container option, a convenience run_container.sh script has been provided to simplify this task:

   # run the sim_cmcl_mods_wrapper container and open its bash terminal
   run_container.sh

Regardless of the docker or local machine option for building and running the wrapper, trigger the following commands to run an example MOO simulations:

   python examples/moo.py
   python examples/moo_analytic.py

Please note that the above examples depend on the mods agent instance being up and running on CMCL servers.

Testing

Testing is done using python's pytest module. The following command would, spin a local mods mock agent, that is included with the tests, and run all the tests with it:

    pytest tests

Contact

For questions, issues, or suggestions, please contact [email protected] and [email protected].