Continuous and Data-Driven Descriptors (CDDD) docker image serving embedding predictions for provided SMILES. The method implementation is found here. Due to numerous dependency errors, and problems running in a local environment, this image provides self-contained environment to transform a list of SMILES into 512-dimensional embeddings.
Build docker image
docker build -t cddd .
Serve in a local environment (by default the fastAPI is exposed on port 80).
docker run --rm -p 80:80 cddd
Use the example file provided.
For unknown smiles model returns nans.
curl -X POST http://localhost:80/predict \
-d @example_batches.json \
-H "Content-Type: application/json"