gw2509 / compounds

Exscientia™ Compounds API

Overview

Web Application

A Java-based web application which provides a CRUD REST API for compounds and their associated assay results. It is backed by an in-memory relational database.

Database

The database schema is generated by the ORM framework (JPA/Hibernate) via annotations in the Java code. The database tables are automatically populated from a SQL population script (src/main/resources/data.sql). Both of the above happen each time the web application starts up. As a consequence, any changes are not persisted between invocations of the web application.

Python (SQL generation) Script

The SQL population script is generated from the provided data/compounds.json and data/schema.json files using a Python script (scripts/json2sql.py). As of this writing, this is done manually, prior to building the application. Note: A pre-generated SQL population script is provided for ease of deployment.

See the Usage section below for build and run instructions.

Notable Features

• Dockerisation (we use Docker™ containers to build and run the web application)
• Integration tests (to test the API's web-service endpoints)
• Swagger (provides a web-based interface to the API for testing and comprehension)
• Pageable (optional endpoint with paging metadata provided for Compound collections)

Usage

Working from the root directory of the project, instructions are as follows:

1. Execute the build script (this may take a few minutes the first time you run it)

2. Execute the run script to start the web application

   ./scripts/build.sh ./scripts/run.sh

You can exercise the web-service endpoints by navigating your browser to: http://localhost:8080/swagger-ui/

Alternatively, you can use the supplied "collection" file (./postman_collection.json) in Postman™ to try it out from there.

Future Improvements

• Data validation on POST and PUT operations
• Implement filtering and ordering features
• Make the API more "hypermedia-driven" by following HATEOAS principles in the API design
• Improve automation by moving database script generation into the Docker™ configuration
• One or two of the (ORM) auto-generated SQL queries are inefficient and should be optimised
• The integration tests are nowhere near complete but hopefully demonstrate the general approach
• The Python script (json2sql.py) is obscenely hard-coded and non-intuitive but served its purpose at the time
• The Swagger API documentation could be prettier and more helpful