scripts/batchtools/README.md

In jmonlong/PopSV: Population-based detection of structural variants from Read-Depth signal

Running PopSV in a HPC using batchtools

Place in the working directory:

• automatedPipeline-batchtools.R which contains the pipeline functions.
• batchtools.conf.R, the configuration file for batchtools.
• batchtools.slurm.tmpl, the template for a job in your HPC.

Then in a R script, call the functions like in run-PopSV-batchtools-automatedPipeline.R (or copy and edit this script). More information on PopSV's website.

To start directly from bin counts (files with columns chr/start/end/bc) follow the commands in run-PopSV-batchtools-automatedPipeline-fromCounts.R.

Configuring your HPC

Slurm

Slurm is used by Compute Canada on Cedar.

batchtools.conf.R is currently set up for Slurm (it calls makeClusterFunctionsSlurm). When running the pipeline functions, this configuration file will be loaded and look for a template named batchtools.slurm.tmpl.

To work on Slurm (or at least on Cedar) you will have to specify your "account". As you can see, there is a field for "account" in the template. You can either replace the <%= resources$account %> by your account ID in the template as you would do when sending jobs manually, or specify your account when calling the R functions later. For example, in run-PopSV-batchtools-automatedPipeline.R, we used other.resources=list(account='rrg-bourqueg-ad') to specify which account we want Cedar to use when sending jobs.

TORQUE

TORQUE is used at McGill's Genome Center on Abacus.

To use TORQUE, batchtools.conf.torque.R should be renamed to batchtools.conf.R in the working directory. This configuration will look for the batchtools.torque.tmpl template file when the pipeline is run.

Others

To use other HPC systems you will need different configuration and template files. There are a few examples of templates on the batchtools GitHub. In the configuration it about which makeClusterFunctions to use (more details in batchtools documentation): makeClusterFunctionsDocker, makeClusterFunctionsInteractive, makeClusterFunctionsLSF, makeClusterFunctionsMulticore, makeClusterFunctionsOpenLava, makeClusterFunctionsSGE, makeClusterFunctionsSSH, makeClusterFunctionsSlurm, makeClusterFunctionsSocket, makeClusterFunctionsTORQUE.

To run the jobs locally across several cores, rename batchtools.conf.local.R to batchtools.conf.R in the working directory. Eventually change the number of cores to use (ncpus= argument in the configuration file).

Testing with a simple job

It's a good idea to check that everything is configured properly before trying to run the pipeline. To test that sending jobs works you could try running the following commands:

library(batchtools)

## To start again from scratch, manually remove the 'test' folder.
reg <- makeRegistry('test', seed=123)
## reg = loadRegistry('test', writeable=TRUE) ## If the registry has already been created before

test.f <- function(ii){
    return(mean(rnorm(10,ii)))
}

batchMap(reg=reg, test.f, 1:2)
submitJobs(reg=reg, ids=findJobs(reg=reg), resources=list(walltime='10:00', cores=1))
waitForJobs(reg=reg, sleep=10)
getStatus(reg=reg)
reduceResultsList(reg=reg)

These commands load the package, create a registry called test, define a function that will be run in the job, setup two jobs with this function and inputs 1 and 2, submit the jobs with a 10min walltime and 1 core per job, wait for the jobs to finish, show a summary of the status, list the results.

jmonlong/PopSV documentation built on Sept. 15, 2019, 9:29 p.m.