openpredict 0.0.5

An API to compute and serve predictions of biomedical concepts associations via OpenAPI for the NCATS Translator project

Translator OpenPredict 🔮🐍 is an API to compute and serve predicted biomedical concepts associations, for the NCATS Translator project.

This service has been built from the fair-workflows/openpredict project.

Use the API 🌐

The Translator OpenPredict API serves predictions of biomedical concepts associations (e.g. disease treated by drug).

Feel free to try it at openpredict.137.120.31.102.nip.io

How to use it

The user provides a drug 💊 or a disease 🦠 identifier as a CURIE (e.g. DRUGBANK:DB00394, OMIM:246300), and choose a prediction model (only OMIM - DrugBank at the moment).

• The /predict call will return the list of predicted target for this entity, the labels are resolved using the Translator Name Resolver API:

{
  "count": 300,
  "relation": "biolink:treated_by",
  "results": [
    {
      "score": 0.8361061489249737,
      "source": {
        "id": "DRUGBANK:DB00394",
        "label": "beclomethasone dipropionate",
        "type": "drug"
      },
      "target": {
        "id": "OMIM:246300",
        "label": "leprosy, susceptibility to, 3",
        "type": "disease"
      }
    }
  ]
}

Try it at https://openpredict.137.120.31.102.nip.io/predict?entity=DRUGBANK:DB00394

• The /query call will return the same predictions in the ReasonerAPI format, based on a ReasonerAPI query
• The /predicates call will return the entities and relations returned by this API (following the ReasonerAPI specifications)

Notebook example

You can find a Jupyter Notebook with examples to query the API on GitHub

Deploy your API 📦

You can also use our tools to build new classifiers, and deploy your OpenPredict API

Install OpenPredict 📥

You might want to use a virtual environment for Python 3.7 to isolate the installation:

# Create the virtual environment in your workspace
python3 -m venv .venv
# Activate it
source .venv/bin/activate

Install the latest release published on PyPI 🏷️ (or see below to run the API with Docker)

pip3 install openpredict

Package available on PyPI: https://pypi.org/project/openpredict

---

Train the model 🚅

Run the pipeline to train the model used by the OpenPredict API.

From a Python script:

from openpredict.openpredict_omim_drugbank import train_drug_disease_classifier

train_drug_disease_classifier()

Or using the command line:

openpredict train-model

Work in progress

model_factory =

---

Run the API ⚙️

The API can be run different ways:

Option 1: Run from the command line

Use the openpredict CLI to start the API using the built classifiers:

openpredict start-api

Access the Swagger UI at http://localhost:8808

Provide the port as arguments:

openpredict start-api --port 8808

Option 2: Run from a Python script

from openpredict import openpredict_api

debug = False
openpredict_api.start_api(8808, debug)

Access the Swagger UI at http://localhost:8808

Run by default in production, set debug = True to run in development environments.

Option 3: Run with Docker

Running using Docker can be convenient if you just want to run the API without installing the package locally, or if it runs in production alongside other services.

Clone the repository:

git clone https://github.com/MaastrichtU-IDS/translator-openpredict.git
cd translator-openpredict

Build and start the openpredict-api container with docker-compose 🐳

docker-compose up -d

Access the Swagger UI at http://localhost:8808

We use nginx-proxy and docker-letsencrypt-nginx-proxy-companion as reverse proxy for HTTP and HTTPS in production. You can change the proxy URL and port via environment variables VIRTUAL_HOST, VIRTUAL_PORT and LETSENCRYPT_HOST in the docker-compose.yml file.

Check the logs:

docker-compose logs

Stop the container:

docker-compose down

---

See also 👀

• Documentation to run in development 📝
• Documentation generated from the source code 📖
• Code of Conduct 🤼