Skip to content

Latest commit

 

History

History
97 lines (63 loc) · 4.37 KB

README.md

File metadata and controls

97 lines (63 loc) · 4.37 KB
Raw

manifest-to-consonance

A tool that converts from a DCC Portal manifest file to Docker container runs using tool images from the Dockstore. It can run these Docker containers locally or it can send "orders" to the Consonance system which runs them on fleets of Cloud VMs.

A use case for this tool is to search for a subset of BAM files from the ICGC PanCancer project indexed on the DCC Portal, select those for analysis with a Docker-based analysis tool from the Dockstore, and then to run the analysis on either a commercial/private cloud or on your local machine. In this way you can do large-scale analysis with the ICGC data files.

You can contrinbute your own Docker-based tools to the Dockstore, see http://www.dockstore.org for more details.

Install

This tool supports two modes, a local mode and a Consonance mode. They each have their own depedencies although the following will need to be setup for any type of run:

  • Perl ** sudo cpanm install YAML::Perl ** sudo apt-get install libjson-pp-perl libconfig-yaml-perl libwww-perl libyaml-perl cpanminus python-pip make libcurl4-openssl-dev python-dev and the same command above since there is no package for YAML::Perl
  • Docker (on Linux) or Boot2Docker (Mac)
  • Java, make sure you have Java 1.8, we have tested 1.8.0_66

Local Mode

You need to have the:

  • Jar file for the tool Launcher available from [here] (https://github.com/CancerCollaboratory/dockstore-descriptor).
    • use 1.0.4 release for the Launcher
    • build with mvn clean install in the launcher directory
    • I also checked in a copy into lib
  • credentials for clouds like AWS depending on where your inputs or outputs are
  • the cwltool command, see here for how to install in detail
    • pip install cwl-runner

Consonance Mode

You need to have the:

  • the consonance command line from here
  • credentials for clouds like AWS, for both inputs/outputs and also launching worker VMs

Collaboratory/AWS Data Download Tool

The ICGC PanCancer data is located on AWS (in S3) and the Collaboratory at OICR. A special tool is required in order to download data from these two sources. You need to download the following to the deps directory.

wget -O deps/dcc-storage-client-0.0.43-dist.tar.gz https://seqwaremaven.oicr.on.ca/artifactory/simple/dcc-release/org/icgc/dcc/dcc-storage-client/0.0.43/dcc-storage-client-0.0.43-dist.tar.gz

And then unzip it.You will also fill in the config file for this tool in deps/dcc-storage-client-0.0.43/conf/application-amazon.properties, make sure you fill in your token:

logging.level.org.icgc.dcc.storage.client=DEBUG
logging.level.org.springframework.web.client.RestTemplate=DEBUG
client.upload.serviceHostname=storage.cancercollaboratory.org
client.ssl.trustStore=classpath:client.jks
accessToken=<your token here>

You then should update the config/launcher.ini file, correct the path below for your system:

[dcc_storage]
client=<path>/deps/dcc-storage-client-0.0.43/bin/dcc-storage-client

Data Location

  • pulling data from S3 requires this tool to run in an AWS VM in Virginia
  • pulling data from Collaboratory requires this tool to run in th Collaboratory OpenStack at OICR

Usage

Note About Shells

The perl script will execute the various commands in a shell started with bash -l -c '<command>' . This means you need to have the following defined in your ~/.bash_profile file:

export JAVA_HOME=<path to JDK 1.8>
export AWS_ACCESS_KEY=<your key>
export AWS_SECRET_KEY=<your secret key>

Local

The tool is called collab since it's for the Collaboratory project.

This will run the BAMStats tool on a collection of BAM files available

perl collab --container-id quay.io/briandoconnor/dockstore-tool-bamstats --manifest manifest.tsv --mode local

Consonance

perl collab --container-id quay.io/briandoconnor/dockstore-tool-bamstats --manifest manifest.tsv --mode consonance

TODO

  • the URL for the dockstore server is hard-coded
  • the downloads are hard-coded to use a 1000 genomes BAM, need to switch to either S3 or Collaboratory-based BAMs
  • the uploads go to an OICR bucket in S3, need to also support (show an example of) upload to Swift provided by Collaboratory