BEL Commons Documentation¶
Installation¶
This application runs on Python 3.7+.
Database¶
For production, it is preferred to use a multi-threading relational database management system. PyBEL has been best tested on PostgreSQL, so this is preferred for now.
Message Broker¶
This application uses Celery as a task management system to support asynchronous parsing of BEL documents, running of analyses, and other slow operations.
RabbitMQ, or any other message queue supported by Celery are appropriate.
Server¶
Because this application is built with Flask, it can be run with the WSGI
protocol. Running on a single machine is possible with either the built-in
werkzeug test server or something easy to install like gunicorn.
For production, uwsgi seems to work pretty well.
Running with Docker¶
Docker is very powerful as a general way to specify how things should be
installed, but has a steep learning curve. After installing and running
docker-machine and docker-compose, BEL Commons can be run with
a few simple commands.
Dockerfile¶
A simple Dockerfile is included at the root-level of the repository. This Dockerfile is inspired by the tutorials from Container Tutorials and Digital Ocean.
Warning
The virtual machine needs at least 2GB memory for the worker container
The database needs a packet size big enough to accommodate large BEL files (>10 mb)
Pages¶
This module contains the user interface blueprint for the application.
Configuration¶
Default configuration can be found in the module bel_commons.config.
By default, PyBEL searches for a configuration file called config.json
in ~/.config/pybel/. This directory can be modified with the environment
variable PYBEL_CONFIG_DIRECTORY. Additioanlly, the location of another
custom configuration can be specified by the environment variable
BEL_COMMONS_CONFIG_JSON.
In config.json add an entry PYBEL_MERGE_SERVER_PREFIX for the address
of the server. Example: http://lisa:5000 with no trailing backslash. This
is necessary since celery has a problem with flask’s url builder function
flask.url_for.
Add an entry PYBEL_CONNECTION with the database connection string to either
a local SQLite database or a proper relational database management system. It’s
suggested to pip install psycopg2-binary in combination with MySQL since it
enables multi-threading.
For a deployment with a local instance of RabbitMQ, the default configuration
already contains a setting for amqp://localhost. Otherwise, an entry
CELERY_BROKER_URL can be set.
-
class
bel_commons.config.BELCommonsConfig(SECRET_KEY: str, BUTLER_PASSWORD: str, BUTLER_EMAIL: str = 'butler', BUTLER_NAME: str = 'BEL Commons Butler', DEBUG: bool = False, TESTING: bool = False, JSONIFY_PRETTYPRINT_REGULAR: bool = True, SQLALCHEMY_TRACK_MODIFICATIONS: bool = False, SQLALCHEMY_DATABASE_URI: str = <factory>, DISALLOW_PRIVATE: bool = True, LOCKDOWN: bool = False, USE_CELERY: bool = True, CELERY_BROKER_URL: str = 'amqp://localhost', CELERY_BACKEND_URL: str = 'redis://localhost', SECURITY_REGISTERABLE: bool = True, SECURITY_CONFIRMABLE: bool = False, SECURITY_SEND_REGISTER_EMAIL: bool = False, SECURITY_RECOVERABLE: bool = False, SECURITY_PASSWORD_HASH: str = 'pbkdf2_sha512', SECURITY_PASSWORD_SALT: str = 'default_please_override', MAIL_SERVER: Optional[str] = None, MAIL_DEFAULT_SENDER_NAME: str = 'BEL Commons', MAIL_DEFAULT_SENDER_EMAIL: Optional[str] = None, REGISTER_EXAMPLES: bool = False, REGISTER_USERS: Optional[str] = None, REGISTER_ADMIN: bool = True, REGISTER_TRANSFORMATIONS: bool = True, WRITE_REPORTS: bool = False, ENABLE_UPLOADER: bool = False, ENABLE_PARSER: bool = False, ENABLE_ANALYSIS: bool = False, ENABLE_CURATION: bool = False, SWAGGER_CONFIG: Mapping[str, Any] = <factory>)[source]¶ Configuration for BEL Commons.
It assumes you have:
SQLite for the PyBEL Cache on localhost
RabbitMQ or another message broker supporting the AMQP protocol running on localhost