solupredict

• download the repositry and
• use "docker compose up" to get started the application

Documentation

File Directory

├── app
│   └─── __init__.py
│   └─── main.py
├── inputs
│   └─── best_model.tar
│   └─── drug_set.json
│   └─── sample_json_data.py
│   └─── test_input.py
│   └─── unwanted_smiles.py
├── model
│   └─── __init__.py
│   └─── load_model.py
│   └─── model.py
│   └─── predict_json.py
├── tests
│   └─── __init__.py
│   └─── test_solubility_json.py
│   └─── test_solubility.py
├── .dockerignore
├── .gitignore
├── Dockerfile
├── entrypoint.sh
├── gunicorn.config.py
├── Procfile
├── requirements.txt
├── README.md

• App folder contains main.py where we handle api calling and posts responses
• inputs folder contains all the inputs required for the model to work
• model folder contains model.py file where all the machine learning code is present. load_model.py contains the code to load the model. If you want to use a custom model load the model in inputs folder and provide the path in load_model.py. predict_json.py contains the code to load the load all the similes in inputs/drug_set.json file and converts them to do predictions.
• tests folder contains the test_solubility_json.py and test_solubility.py to do unit tests
• Dockerfile used to deploy this in a container
• entrypoint.sh conatins the starting command to start the fastapi in docker after deploying and run background workers
• requirements.txt contains all the required modules for this repository