CryptoShow

CryptoShow is an application for detection of protein binding sites utilizing the CryptoBench model and ESM-2 embeddings.

Build

Build using Docker: docker compose --profile <cpu/gpu> build or docker buildx bake <cpu/gpu> (experimental). Make sure to specify --profile gpu to run with CUDA or --profile cpu to run on CPU.

For monitoring, also include --profile monitoring.

Run

To run the Docker Compose, use docker compose --profile <cpu/gpu> up.

The frontend is available at localhost:80.

For SSL, set ENABLE_SSL=true, CERTBOT_ENABLED=true, DOMAIN and CERTBOT_EMAIL in your .env and add --profile certbot when starting. Certbot will automatically obtain and renew Let's Encrypt certificates inside Docker.

Also, when developing behind a proxy, change the HTTP_PROXY and HTTPS_PROXY env variables.

Local Frontend Development

See frontend/README.md.

Local Backend Development

See backend/README.md.

Deployment

1. Install Docker
2. Create .env and set up the vars (for SSL, see the instructions above)
3. Enable ports in firewall (sudo ufw allow 443, sudo ufw allow 80)
4. Set up the permissions for the user 2727, i.e. sudo chown -R 2727:2727 ./data. Optionally, customize the UID and GID env vars.
5. Run Docker
6. Optionally, set up the continuous deployment (see .github/workflows/production-deploy.yml, make sure that the user is in the docker group and that sudo chown root:docker /var/run/docker.sock && sudo chmod 660 /var/run/docker.sock)
7. Optionally, set up monitoring credentials by creating the .htpasswd file (sudo apt install apache2-utils, htpasswd -c ./frontend/monitoring.htpasswd admin)
8. If you run on Linux, make sure ./data/monitoring/grafana and ./data/monitoring/prometheus are writable by Docker.

Technical Requirements

• RAM: minimum 32 GB, preferred 64 GB
• CPU: minimum 8 cores, preferred 8+ cores
• OS: anything running Docker (preferrably Linux)
• disk: depends on number of processed structures

Maintenance Mode

To toggle the maintenance mode, create/delete the ./frontend/maintenance/maintenance.flag file (the file can be empty).

Monitoring

After setting up the monitoring credentials (see Deployment), you can access the monitoring services via localhost/grafana and localhost/flower. Grafana dashboards, users, and Prometheus metrics are stored under ./data/monitoring, so they persist across normal redeploys.

TLDR

docker-compose --profile <gpu/cpu> up --build, localhost:80/443