This package provides a set of binaries (and a Perl module) to interact with the NBI HPC cluster.

List or delete your jobs in the queue with lsjobs. By default search is restricted to the active user, which can be changed with -u USER or set to -u all.

See docs

lsjobs [options] [jobid.. | pattern ]

Add -d to delete the jobs (you will be prompted for confirmation).

Submit a job to the queue (with cores, memory, time, etc) with runjob.

See docs

runjob -n "my-job" -t 2 -r -c 18 -m 32 --after 9290 "python script.py --threads 18"

If invoked without --run (or -r) it will print the command to be executed.

Wait for all jobs matching a pattern to finish (to be used to run a second job when they are all finished), with waitjobs. You can send it to the queue, and use its JOBID as dependency for other jobs.

See docs

waitjobs [-u $USER] [pattern]

List packages installed in the NBI HPC with shelf. See docs

shelf QUERY_STRING

Generate a Singularity image from a bioconda package with make_image_from_bioconda.

See docs

make_image_from_bioconda --output seqfu-1.20.sif --version 1.20.0 --package seqfu

Given a singularity image, "install" a package in the NBI HPC (to be used with source package xyz)

See docs

make_package --image seqfu-1.20.sif --package seqfu__1.20   seqfu fu-orf fu-rename fu-msa...

⚠️ If you generated the image with make_image_from_bioconda you should be able to omit the list of binaries, as they are stored in the image itself.

Start an interactive session with session, defaults can be saved in the configuration file.

See docs

session [-t 2 -c 18 -m 32]

Who is using the cluster: list usernames and number of jobs in ascending order with whojobs. See docs

whojobs [--min-jobs INT]

The scripts will look for a configuration file in ~/.nbislurm.config where you can store your email address, default queue, and other options, as well as default parameters for interactive sessions. See the docs.

NBI::Slurm is a Perl package that provides a convenient interface for submitting jobs to SLURM, a workload manager for High-Performance Computing (HPC) clusters. It includes two main classes to submit jobs to SLURM:

• NBI::Job, which represents a job to be submitted to SLURM, and
• NBI::Opts, which represents the SLURM options for a job.

And two classes to manage the output of the jobs:

• NBI::Queue, which represents the content of the SLURM queue, and
• NBI::QueuedJob, which represents a single job in the queue.

Features

• Very experimental, very alpha, very buggy.
• Easily create and configure SLURM jobs using the NBI::Job class.
• Comes with binaries to submit jobs, list jobs, wait for jobs, etc.

To use NBI::Slurm, you need Perl 5.12 or higher installed on your system. You can install the package using CPAN or manually by copying the NBI/Slurm.pm file to your Perl library directory.

cpanm is a command line utility for installing Perl modules from CPAN.

To install cpanm, run the following command:

# If you dont have cpanm installed:
curl -L https://cpanmin.us | perl - --sudo App::cpanminus

# Install with cpanm
cpanm NBI::Slurm

Here's a simple example demonstrating how to use NBI::Slurm to submit a job to SLURM:

use NBI::Job;
use NBI::Opts;

# Create a job
my $job = NBI::Job->new(
    -name    => "job-name",
    -command => "ls -l",
);

# Create options
my $opts = NBI::Opts->new(
    -queue   => "short",
    -threads => 4,
    -memory  => 8,
    -opts    => ["-m afterok:1234"],
);

# Set options for the job
$job->set_opts($opts);

# Submit the job to SLURM
my $jobid = $job->run;

For more detailed information on the available methods and options, please refer to the individual documentation of the NBI::Job and NBI::Opts classes.

NBI::Slurm is written by Andrea Telatin

This module is released under the MIT License.