m2m - Man to Molecule to Man

Artificially intelligent drug repurposing over federated knowledge sources.

This is a first checkin to start the conversation about the best structure for this system.

Structure

The structure below provides a root m2m package with modules for:

• App The application can be run from the command line.
• Hypothesis and Evaluation Inputs are repurposing hypotheses. Outputs are evaluations of these hypotheses.
• Knowledge: This is a source for data which repurposing algorithms evaluate.
• Algorithm: A module for aggregating various evaluation methods.
• Interface: We might have several interfaces. Interfaces generate repurposing hypotheses.
• API: A web based smartAPI for the repurposing functionality. Multiple user interfaces might use this service.
• Similarity: A store for a library of similarity metrics.

├── README.md
├── api
│   └── service.py
└── m2m
    ├── app.py
    ├── evaluation.py
    ├── hypothesis.py
    ├── knowledge.py
    ├── repurpose
    │   ├── __init__.py
    │   ├── algorithm
    │   │   ├── __init__.py
    │   │   ├── core.py
    │   │   └── literature.py
    │   ├── api
    │   │   └── server.py
    │   └── interface
    │       ├── __init__.py
    │       ├── cli.py
    │       └── core.py
    └── similarity
        └── pmc.py

To Run

Setup

Install Python 3.6.x Clone the repo. Install requirements from the requirements.txt file.

The application

This will create trivial interfaces and algorithms, wire them together, and output an evaluation.

[scox@mac~/dev/m2m]$ PYTHONPATH=$PWD python m2m/app.py

The Similarity API

[~/dev/m2m]$ PYTHONPATH=$PWD python m2m/api/service.py

The service will be available at http://localhost:5000/apidocs.

The API

This will start a web API.

[~/dev/m2m]$ PYTHONPATH=$PWD python m2m/repurpose/api/server.py

At http://localhost:5000/apidocs, you should get this: