BioKb-ChEBI

BioKb-ChEBI (biokb_chebi) is a python package to import ChEBI data into a relational database and create RDF triples (turtles) from it. The turtles can be imported into a Neo4J graph database. The package is part of the BioKb family of packages to create and connect biological and medical knowledge bases and graphs.

The package provides different options to run it: from command line, as RESTful API server, as Podman/Docker container, or as Podman/Docker networked containers with Neo4J and a relational database.

Features

biokb_chebi allows to ...

1. Query ChEBI data with SQLAlchemy or raw SQL
2. Load, query and manage ChEBI data with GUIs for knowledge base and graphs (phpMyAdmin, Neo4J Browser)
3. Query data via a RESTful API (FastAPI) with OpenAPI documentation and interactive Swagger-UI

to provide this biokb_chebi ...

• imports ChEBI data into a relational database
• creates RDF triples (turtles) from the relational database
• imports the RDF triples into a Neo4J graph database

Supported databases: SQLite, MariaDB/MySQL, PostgreSQL, Oracle, Microsoft SQL Server, and any other database supported by SQLAlchemy.

Options to run BioKb-ChEBI

All biokb packages share the same API and CLI structure. You have different options to run the packages:

1. from command line (simplest way to get started)
2. as RESTful API server (can start directly from command line)
3. as Podman/Docker container (without import into Neo4J, but export of turtles possible)
4. as Podman/Docker networked containers (with all features) and 3 containers:
   1. high-performance relational databases (PostgreSQL, Oracle, MySQL, ...)
   2. RESTful API (fastAPI) for queries, data import and export
   3. GUI for querying and administration of MySQL over the Web

Installation

If uv is installed:

uv venv
source .venv/bin/activate
uv pip install biokb_chebi

Otherwise:

python3 -m venv .venv
source .venv/bin/activate
pip install biokb_chebi

Run BioKb-ChEBI

From command line

For sure the simplest way is to run all steps:

biokb_chebi import-data
biokb_chebi create-ttls

Before importing into Neo4J, make sure Neo4J is running (see below "How to run Neo4J").

Then import into Neo4J:

biokb_chebi import-neo4j -p neo4j_password

http://localhost:7474 (user/password: neo4j/neo4j_password)

For more options see the CLI options section below.

As RESTful API server

Usage: biokb_chebi run-api [OPTIONS]

biokb_chebi run-api

• user: admin
• password: admin

http://localhost:8000/docs#/

1. Import data
2. Export ttls
3. Run Neo4J (see below "How to run Neo4J")
4. Import Neo4J

Be patient, each step takes several minutes.

As Podman/Docker container

For docker just replace podman with docker in the commands below.

Build & run with Podman:

git clone https://github.com/biokb/biokb_chebi.git
cd biokb_chebi
podman build -t biokb_chebi_image .
podman run -d --rm --name biokb_chebi_simple -p 8000:8000 biokb_chebi_image

• Login: admin
• Password: admin

With environment variable for user and password for more security:

podman run -d --rm --name biokb_chebi_simple -p 8000:8000 -e API_PASSWORD=your_secure_password -e API_USER=your_secure_user biokb_chebi_image

http://localhost:8000/docs

On the website:

1. Import data
2. Export ttls

Neo4j import in this context is not possible because Neo4J is not running in the same network as service, but the exported turtles can be imported into any Neo4J instance using the CLI (biokb_chebi import-neo4j).

to stop the container:

podman stop biokb_chebi_simple

to rerun the container:

podman start biokb_chebi_simple

Run as Podman/Docker networked containers

If you have docker or podman on your system, the easiest way to run all components (relational database, RESTful API server, phpMyAdmin GUI) is to use networked containers with podman-compose/docker-compose.

git clone https://github.com/biokb/biokb_taxtree.git
cd biokb_taxtree
podman-compose -f docker-compose.db_neo.yml --env-file .env_template up -d
podman-compose --env-file .env_template up -d

http://localhost:8001/docs

On the website:

1. Import data
2. Export ttls
3. Import Neo4J

stop with:

podman pod stop pod_biokb_db
podman-compose stop

rerun with:

podman pod start pod_biokb_db
podman-compose start

Tip: Copy the .env_template to .env and change the default passwords in the .env file before starting the containers for better security. If you have done that you need to use --env-file .env instead of --env-file .env_template in the commands above or just omit the --env-file option (because the default is .env).

CLI Options

Import data into relational database

Usage: biokb_chebi import-data [OPTIONS]

biokb_chebi import-data

-> SQLite database in ~/.biokb/biokb.db. Open with e.g. DB Browser for SQLite

If you want to use different relational database (MySQL, PostgreSQL, etc.), provide the connection string with -c option. Examples:

• MySQL: mysql+pymysql://user:password@localhost/biokb
• PostgreSQL: postgresql+psycopg2://user:password@localhost/biokb

For more examples please check how to create database URLs

Create RDF turtles

Usage: biokb_chebi create-ttls [OPTIONS]

biokb_chebi create-ttls

-> RDF turtles will be created in ~/.biokb/chebi/data/ttls.zip

Import into Neo4J

Start Neo4J ...

podman run --rm --name biokb-neo4j-test -p7474:7474 -p7687:7687 -e NEO4J_AUTH=neo4j/neo4j_password neo4j:latest

Note: Remove --rm if you want to keep the container after stopping it. Replace podman with docker if you use Docker.

... and import into Neo4J:

biokb_chebi import-neo4j -p neo4j_password

http://localhost:7474 (user/password: neo4j/neo4j_password)

How to run Neo4J

For the options "Run BioKb-ChEBI as ..."

1. From command line
2. As RESTful API server

you need to run Neo4J separately.

If you have not already a Neo4j instance running, the easiest way is to run Neo4J as Podman/ Docker container.

For docker just replace podman with docker in the commands below.

podman run -d --rm --name biokb-neo4j -p7474:7474 -p7687:7687 -e NEO4J_AUTH=neo4j/neo4j_password neo4j:latest
# Remove `--rm` if you want to keep the container after stopping it.

Neo4J is then available at: http://localhost:7474 (user/password: neo4j/neo4j_password

Stop Neo4J with:

podman stop biokb-neo4j

if you have not used --rm above, you can restart Neo4J with:

podman start biokb-neo4j

Query database with SQLAlchemy

In order to query the database with SQLAlchemy you can use the following code snippet:

import os
from biokb_chebi import get_session
from biokb_chebi.db.models import Compound
os.environ.pop("CONNECTION_STR", None)  # to make sure no environment variable is used
with get_session() as session:
    results = session.query(Compound).filter(Compound.name.ilike("%glucose%")).limit(2)
    for row in results:
        print(f"Name: {row.ascii_name}, Name: {row.source}")

Output:

Name: 1,2,3,4-tetrakis-O-galloyl-alpha-D-glucose, Name: KEGG COMPOUND
Name: 1-caffeoyl-beta-D-glucose, Name: KEGG COMPOUND