Skip to content
This repository was archived by the owner on Apr 25, 2018. It is now read-only.

Latest commit

 

History

History
99 lines (81 loc) · 3.81 KB

toil-rnaseq-unc.sop.md

File metadata and controls

99 lines (81 loc) · 3.81 KB

Steps to run the RNA-seq UNC pipeline vs the RNA-seq CGL pipeline

Setup the cluster the same way, but instead of needing a venv and pip installing toil-scripts, git clone toil-scripts -b releases/2.0.x you'll also need to use cgcloud ssh-cluster to pip uninstall toil and reinstall pip install toil==3.1.6 to all the nodes

There is a convenience script for running on mesos, which you'll edit to change bucket other settings etc. The config you create (there is an example in the directory), must be on every machine, so you'll use the --share option and pass in a location where that file is, which will be on /home/mesosbox/shared You will also need to have the master key file in the shared directory

  1. Setup CGCloud See the SOP toil-rnaseq.sop.md
  2. Launch a cluster See the SOP toil-rnaseq.sop.md

Instead of section 3 and 4 in the toil-rnaseq.sop.md use the following:

  1. Install and configure the pipeline #ssh as administrator into the Toil leader (-a) so you have sudo rights cgcloud ssh —a —-cluster-name example toil-leader #install git on the leader sudo apt-get install git #exit out of the leader exit

#grow the cluster #use an appropriate spot bid price cgcloud grow-cluster --cluster-name example --spot-auto-zone --spot-bid 1.00 --instance-type c3.8xlarge --num-workers 22 toil

#remove current Toil from the leader and the workers #use CGCloud to remove the current Toil version from the workers cgcloud ssh-cluster --cluster-name example --admin toil -o StrictHostKeyChecking=no sudo pip uninstall -y toil

#install Toil version 3.1.6 on all the workers cgcloud ssh-cluster --cluster-name example --admin toil -o StrictHostKeyChecking=no sudo pip install toil==3.1.6

#install unzip and samtools; these are not installed by default cgcloud ssh-cluster --cluster-name example --admin toil -o StrictHostKeyChecking=no sudo apt-get install -y unzip samtools

#log back into the Toil leader cgcloud ssh --cluster-name example toil-leader

#clone the correct toil scripts tag git clone https://github.com/BD2KGenomics/toil-scripts -b releases/2.0.x

#edit the launch script to make sure paths and options are setup correctly #especially the location of the config file and the S3 output directory vim /home/mesosbox/toil-scripts/src/toil_scripts/rnaseq_unc/launch_unc_mesos.sh #for example 'aws:us-west-2:unc-run-1 ' -> 'aws:us-west-2:example-unc-run-1 ' '--s3_dir cgl-driver-projects/test/ ' -> '--s3_dir cgl-driver-projects-encrypted/wcdt/issue1111/output/ '

#cd to the scripts directory cd /home/mesosbox/toil-scripts/src/toil_scripts/rnaseq_unc #set the script as executable chmod 775 launch_unc_mesos.sh

#turn off ‘Requester Pays’ on the Amazon S3 console so that downloads can occur #(turn this back on after the run finishes downloading files) go to the S3 console https://console.aws.amazon.com/s3/home?region=us-west-2 click on All Buckets click on cgl-pipline-inputs click on Properties click on Requester Pays uncheck the ‘Enable’ box click ’Save’

  1. Run the pipeline #start a screen session so we can log back in and see progress screen -S myRNASeq-UNC-session

#run the launch script ./launch_unc_mesos.sh

#if a problem occurs and you need to restart #add '--restart' to the end of the launch_unc_mesos.sh script and rerun the script #or #if you need to clean up all of the previous run #do the following to clean up first (change to the appropriate region and run name) toil clean 'aws:us-west-2:example-unc-run-1' #and rerun the launch_unc_mesos.sh script

#after the run finishes or after downloads have occurred #turn on ‘Requester Pays’ on the Amazon S3 console so that downloads can occur go to the S3 console https://console.aws.amazon.com/s3/home?region=us-west-2 click on All Buckets click on cgl-pipline-inputs click on Properties click on Requester Pays check the ‘Enable’ box click ’Save’