|
| 1 | +**weave** is capable of automatically distributing its pipeline jobs across a slurm cluster. The context for it's initial execution can be varied as well. |
| 2 | + |
| 3 | +The context is also centrally related to the configuration and setup of a particular cluster. Right now weave is configured to work with NIH clusters **skyline**, **biowulf**, and **bigsky**. |
| 4 | + |
| 5 | + |
| 6 | +Typical contexts of execution include: |
| 7 | + |
| 8 | +# srun (real time execution) (non-interactive) |
| 9 | + |
| 10 | +The **weave** pipeline can be triggered from a head node in a non-interactive fashion: |
| 11 | + |
| 12 | +## Bigsky/Skyline |
| 13 | + |
| 14 | +!!! Note |
| 15 | + Dependency files for skyline and bigsky differ |
| 16 | + Bigsky: `/gs1/RTS/OpenOmics/bin/dependencies.sh` |
| 17 | + Skyline: `/data/openomics/bin/dependencies.sh` |
| 18 | + |
| 19 | +```bash |
| 20 | +source ${dependencies} |
| 21 | +srun --export=ALL "weave run [keyword args] ${run_id}" |
| 22 | +``` |
| 23 | + |
| 24 | +!!! Note |
| 25 | + srun (by default)[https://slurm.schedmd.com/srun.html#OPT_export] exports all environmental variables from the executing environment and `--export=ALL` can be left off |
| 26 | + |
| 27 | +## Biowulf |
| 28 | + |
| 29 | +```bash |
| 30 | +srun --export=ALL "module load snakemake singularity; weave run [keyword args] ${run_id}" |
| 31 | +``` |
| 32 | + |
| 33 | +# srun (real time execution) (interactive) |
| 34 | + |
| 35 | +## Bigsky/Skyline |
| 36 | + |
| 37 | +!!! Note |
| 38 | + Dependency files for skyline and bigsky differ |
| 39 | + Bigsky: `/gs1/RTS/OpenOmics/bin/dependencies.sh` |
| 40 | + Skyline: `/data/openomics/bin/dependencies.sh` |
| 41 | + |
| 42 | +```bash |
| 43 | +> # <head node> |
| 44 | +srun --pty bash |
| 45 | +> # <compute node> |
| 46 | +source ${dependencies} |
| 47 | +weave run [keyword args] ${run_id} |
| 48 | +``` |
| 49 | + |
| 50 | +## Biowulf |
| 51 | + |
| 52 | +```bash |
| 53 | +> # <head node> |
| 54 | +sinteractive |
| 55 | +> # <compute node> |
| 56 | +module purge |
| 57 | +module load snakemake singularity |
| 58 | +weave run [keyword args] ${run_id} |
| 59 | +``` |
| 60 | + |
| 61 | +Biowulf uses environmental modules to control software. After executing the above you should see a message similar to: |
| 62 | + |
| 63 | +> [+] Loading snakemake 7.XX.X on cnXXXX<br /> |
| 64 | +> [+] Loading singularity 4.X.X on cnXXXX<br /> |
| 65 | +
|
| 66 | +# sbatch (later time execution) |
| 67 | + |
| 68 | +## Bigsky/Skyline |
| 69 | + |
| 70 | +### sbatch tempalte |
| 71 | +```bash title="<b>bigsky-skyline sbatch template</b>" |
| 72 | +#!/bin/bash |
| 73 | +#SBATCH --job-name=<job_name> |
| 74 | +#SBATCH --export=ALL |
| 75 | +#SBATCH --time=01-00:00:00 |
| 76 | +#SBATCH --cpus-per-task=1 |
| 77 | +#SBATCH --ntasks=1 |
| 78 | +#SBATCH --mem=8g |
| 79 | +#SBATCH --output=<stdout_file>_%j.out |
| 80 | +source ${dependencies} |
| 81 | +weave run \ |
| 82 | +-s /sequencing/root/dir \ |
| 83 | +-o output_dir \ |
| 84 | +<run_id> |
| 85 | +``` |
| 86 | + |
| 87 | +This above script can serve as a template to create an sbatch script for weave. Update the psuedo-variables in the script to suit your particular needs then execute using sbatch command: |
| 88 | + |
| 89 | +```bash |
| 90 | +sbatch weave_script.sbatch |
| 91 | +``` |
| 92 | + |
| 93 | +## Biowulf |
| 94 | + |
| 95 | +### sbatch tempalte |
| 96 | +```bash title="<b>biowulf sbatch template</b>" |
| 97 | +#!/bin/bash |
| 98 | +#SBATCH --job-name=<job_name> |
| 99 | +#SBATCH --export=ALL |
| 100 | +#SBATCH --time=01-00:00:00 |
| 101 | +#SBATCH --cpus-per-task=1 |
| 102 | +#SBATCH --ntasks=1 |
| 103 | +#SBATCH --mem=8g |
| 104 | +#SBATCH --output=<stdout_file>_%j.out |
| 105 | +module purge |
| 106 | +module load snakemake singularity |
| 107 | +weave run \ |
| 108 | +-s /sequencing/root/dir \ |
| 109 | +-o output_dir \ |
| 110 | +<run_id> |
| 111 | +``` |
0 commit comments