PTP Project

Overview

This project is designed to process and analyze molecular data using Ligand based models run with CPSign. It includes components for data storage, processing, and web-based interaction.

Running the project

1. Navigate to the specified serving url.
2. Upload a csv smiles file.

Se the example in test/smiles_example.csv for valid input file format.

Local Installation

There are two options. Docker-compose is fast to get started. The alternative and production-like alternative is skaffold. See skaffold.dev for more information on how to run.

Usage

To use this project, follow these steps:

1. Clone the repository:

   git clone https://github.com/pharmbio/ptp.git
   cd ptp

2. Run the application:

   docker-compose up --build

Deployment

Kubernetes Deployment

1. Modify values.yaml according to your specification.

2. Deploy using Helm:

   helm install --upgrade charts/ptp --values values.yaml

Note: The deployment will create a Job to fetch all depdendency models before starting services. This may take some time since there are 10Gb+ downloadables. This is only needed on load (new version) or reload of model library.

Docker Deployment

1. Build and run the Docker containers:

   docker-compose up --build

Scilifelab Serve Deployment

1. Build container

   docker build -t ptp .

2. Push container to registry

3. Deploy

   • By default the folder /app/ext_storage/ should be mounted to outside storage to avoid ram bloat. It will contain database, media, and models.
   • Place the downloaded models (from https://huggingface.co/pharmbio/ptp) to /app/ext_storage/models/

4. Configuration

   • Currently static files are served by Django itself even with DEBUG=False
   • All defaults of environmental variables make sense for SciLifeLab Serve. Here are some environmental variables that can be changed if needed:
   • EMAIL_HOST, EMAIL_PORT, EMAIL_HOST_USER etc for email settings.
   • MAX_MODELS to limit the number of active models (for debug purposes) .otherwise it iterates all models present (currently 800+).
   • MEDIA_DIR for media directory
   • MODEL_DIR for models directory
   • DATABASE_DIR for database directory
   • SITE_URL for generating download links in emails

References

Model repository on Hugging Face: pharmbio/ptp

License

This project is licensed under the MIT License. See the LICENSE file for details.

Authors

• Jonathan Alvarsson - Research, scipipe version and models - jonathanalvarsson
• Morgan Ekmefjord - Web service, deployment and service packaging - morganekmefjord