Spaces:
Running
Running
ENCODE application source
The current architecture, model conventions, data layout, and deployment steps are documented once in the repository README.
Layout
backend/— FastAPI routes, BGE-FT-VA query encoding, FAISS code search, two-surface phenotype retrieval, mappings, and graph responses.frontend/— static HTML, CSS, and JavaScript served by the backend and reused by the Netlify package.requirements.txt— pinned CPU runtime dependencies used by the Docker image.
For a local source-mode server with dependencies already installed:
uvicorn src.backend.app:app --host 127.0.0.1 --port 8000
The required model, indexes, embeddings, and runtime sidecars must exist under
outputs/ or be redirected through the ENCODE_* environment variables defined
in backend/paths.py.