Skip to content

galaxy-genome-annotation/docker-chado

Repository files navigation

Docker Chado Container Docker Repository on Quay DOI

Built on top of a standard postgres:16 container, the Chado container provides the Chado schema loaded with all 5 standard ontologies.

Launching the Container

The Chado container is very simple to start, as simple as a regular postgres container:

docker run -d --name chado quay.io/galaxy-genome-annotation/chado:latest

The Chado schema in this container will not persist across restarts, to allow for that, supply a volume with -v like so:

docker run -d --name chado -v /path/to/storage:/var/lib/postgresql/data/ quay.io/galaxy-genome-annotation/chado

The schema and five default ontologies are installed upon launch, if no chado instance is detected.

Accessing the Container

If you haven't exposed a port with the -p/-P options, you can connect to your chado container via a linked container:

docker run -i -t --link chado:db quay.io/galaxy-genome-annotation/chado bash
root@0069babbd55f:/# psql -h db -U postgres postgres
# Password is postgres

Schema Free Container

If you do not wish to have the schema automatically installed upon launch, simply set the environment variable INSTALL_CHADO_SCHEMA to 0:

docker run -d --name chado-tools -e INSTALL_CHADO_SCHEMA=0 quay.io/galaxy-genome-annotation/chado

This will let you use all the GMOD tools without needing to wait for the chado schema to install.

Yeasty Container

If you would like some default yeast data installed for you, you can supply the environment variable INSTALL_YEAST_DATA=1. This requires that you leave INSTALL_CHADO_SCHEMA=1:

docker run -d --name chado-yeast -e INSTALL_YEAST_DATA=1 quay.io/galaxy-genome-annotation/chado

Using the Container in docker-compose.yml

It is strongly, strongly recommended that you pin your images to a specific tag of this repository. I have unintentionally broken the :latest images before.

E.g.

image: quay.io/galaxy-genome-annotation/chado:1.31-jenkins61-pg9.5

Given that we, as the developers, have no easy way to communicate to you as the end user that breaking changes have been made (and keeping backwards compatability is prohibitve for a small team..., sorry!), it is best to pin and read the changelog before upgrading.

Changelog

  • 2024-02-22
    • @hexylena updated postgres from 9 to 16
    • @hexylena rebuilt with latest CSB data which is still quite old.
  • 2021-04-01
    • @mboudet fixed the missing yeast genome
  • 2017-02-21
    • Re-arranged image to decrease layers.
    • Added some custom SQL required for postgraphql hacks.
  • [sometime recently]
    • changed PGDATA to match the current upstream value (/var/lib/postgresql/data/) rather than the version specific directory.