website-jbrowse-gff

Dockerfile for processing GFF data for JBrowse

This README encompasses not only what is contained in this repo but also documents how it fits in with the Alliance of Genome Resources data and server pipelines, as well as the the data processing and JBrowse server tools in the website-genome-browsers repo in the jbrowse-* branches.

Overview

The Dockerfile in this repo codes for a data processing tool that fetches GFF from the WormBase FTP site, processes it into JBrowse NCList json format, generates search index files, generates "preliminary" trackList.json files for each assembly (indicating what data types are available for each), and deposits the results in the Alliance JBrowse S3 bucket. This processing is currently configured to run through the Alliance GoCD system and making use of the ridiculously parallelizable nature of processing data from many assemblies. If running on a single CPU is desired, the single_species branch can be used (it is somewhat misleadingly named, as it will run all species, but one at a time).

Workflow

When starting a new release, a release-specific branch is created from the jbrowse-staging branch, typically called jbrowse-$RELEASE. Usually, only two changes need to be made for a release: A) bump the RELEASE= line in /website-genome-browsers/jbrowse/Dockerfile, and B) bump the release number in /website-genome-browsers/jbrowse/plugins/wormbase-glyphs/js/main.js (look for "WS$REELASE" in this file to find what needs to be updated). Once the release number has been pushed into the release specific repo in website-genome-browser, two changes need to be made in this repo:

1. The Dockerfile in this repo should be updated to change the value of the RELEASE ARG to the current release number:

ARG RELEASE=293

2. A similar line in the parallel.sh script should also be changed:

RELEASE=293

Both of these changes should be committed back to the main branch of this repo.

Note that both of these items could potentially be parameterized and passed in when running though Ansible using the $MOD_RELEASE_VERSION environment variable in the agr_ansible_devops WormBase specific environment (https://github.com/alliance-genome/agr_ansible_devops/blob/master/environments/jbrowse/wb.yml), but this hasn't been hooked up yet.

When these changes are commited to the main branch, the GoCD system will run the JBrowseSoftwareProcessWB pipeline to build the Dockerfile in this repo, and then run the JBrowseProcessWB pipeline, which will run a compute machine through Ansible to process the WormBase GFF files. The script that it runs, parallel.sh uses GNU parallel to process all of the assemblies in WormBase (currently 31). After processing the the files, the script will upload the JBrowse data to the Alliance JBrowse S3 bucket (agrjbrowse).

IMPORTANT NOTE about GoCD: typically, the JBrowseSoftwareProcessWB and JBrowseProcessWB are paused to prevent them from accidentally running when updates to this repo are commited. Be sure to unpause them when you want these pipelines to run.

Building JBrowse servers

See the documentation in https://github.com/WormBase/website-genome-browsers/blob/master/README.md for build procedures for the AWS Amplify-powered JBrowse instances.