BIRCH on HPC systems

From Bioinformatics.Org Wiki

Jump to: navigation, search

Mystica Arrow set (with deep) 1.png [return to Release To Do List]


Rationale: HPC systems are generally pretty opaque about what is on the system and how to use tools. Nowhere is the user-oriented advantages of BIRCH needed more than on multiuser HPC systems.

While BIRCH has evolved from day 1 as a multiuser resource, there are certainly going to be challenges on major platforms that will need to be addressed. This page will be the place for making BIRCH the best solution for a user accessible HPC system.

The test platform will be the Digital Alliance of Canada.

Important DCA links

Contents

About DCA

The nibi.alliancecan.ca runs AlmaLinux 9.6, which is a free distro of RedHat Linux.

Getting BIRCH to run on DAC

Unique aspects of the DAC system

module load module_name

module load only applies to the shell in which it was launched, so if you want a personal default, the user has to run module load in their .profile, .bashrc other file at login.

Example of lines to add to a SLURM job:

#SBATCH --mail-user=your.email@example.com
#SBATCH --mail-type=ALL

The Slurm documentation explains the possible values for the `--mail-type` option: https://slurm.schedmd.com/sbatch.html#OPT_mail-type

Java

in local.profile.source

module load java

To load the default java version each time a shell is launched. I tried putting this line in local.profile.login.source, but it appeared to have no effect.

Personal tools
Namespaces
Variants
Actions
wiki navigation
Toolbox