GitHub Project Planning

This repo implements a BioCypher pipeline that grabs all issues on a GitHub Projects board that simulates a collaborative team (available here). The pipeline builds a knowledge graph of issues, categories, assigned team members, and annotations on the issues, and then provides access to an LLM assistant that summarises tasks performed in the previous iteration of the project for the group and each individual. Further, the web app provides a planning tab that analyses tasks in next iterations of the project and provides suggestions for task prioritisation and collaboration, again for the group and each individual.

Usage

Important

The knowledge graph build stage requires a GitHub token to access the GitHub API, BIOCYPHER_GITHUB_PROJECT_TOKEN, in the environment. If you don't have access to this token, the KG build will fail. You can see an online version of the app here.

The docker-compose.yml file contains the necessary services to run the pipeline. To start the pipeline, run:

git clone https://github.com/biocypher/project-planning.git
cd project-planning
docker-compose up -d

The pipeline will build, import, and deploy the knowledge graph, and then start the app service. The KG is available at http://localhost:7474 and the app is available at http://localhost:8051.

This standard pipeline runs a non-password-protected Neo4j instance. To run a password-protected instance (with ports open to the web), run:

docker-compose -f docker-compose-password.yml up -d

Authentification settings are configured in the docker-compose-password.yml

Knowledge Graph

The build procedure is orchestrated by BioCypher. The central script is create_knowledge_graph.py, which is responsible for calling the adapter to fetch data from the GitHub API and building the KG. You can run it locally with:

poetry install
poetry run python create_knowledge_graph.py