Skip to contents

NOAA SEDNA setup

Source: vignettes/NOAA-SEDNA.Rmd
NOAA-SEDNA.Rmd

This page covers what is specific to the NOAA NMFS SEDNA cluster. Once you are set up here, RiboPilot works as in the Get started walkthrough, with executor = "NOAA_SEDNA". Running on a cluster explains how RiboPilot uses a cluster in general.

You need a SEDNA account; see the account instructions, or contact Krista Nichols ().

First-time setup

RiboPilot needs Nextflow, Singularity, and R. On SEDNA, put all three in one Mamba environment; installing RiboPilot with the cluster’s R module runs into problems.

Log in to the NOAA NMFS VPN, then to SEDNA. If you have never used Mamba on SEDNA, run this once:

/opt/bioinformatics/mambaforge/bin/mamba init

Create the environment. This may take a while.

mamba create --name RiboPilot_deps 'bioconda::nextflow>=24.10,<25.11' \
  conda-forge::singularity conda-forge::zlib 'conda-forge::r-base>=4.4' -y

Then install RiboPilot into it:

mamba activate RiboPilot_deps
R

In that R session, follow step 5 of Install the tools, then leave R with quit().

To upgrade later, activate RiboPilot_deps first, then follow Upgrading RiboPilot.

Open the app

The app needs little computing power, since the heavy work goes to the queue. Start a small interactive session, note the name of the node it gives you, and start R:

srun -c 2 --mem=16GB -p standard --pty /bin/bash
hostname
mamba activate RiboPilot_deps
R

In R, create a project (once), then open the app without a browser:

RiboPilot::new_test_project(
  path = "/path/to/RiboPilot/test_project",
  executor = "NOAA_SEDNA"
)
RiboPilot::RiboPilot("/path/to/RiboPilot/test_project",
  host = "0.0.0.0", port = 8123, launch.browser = FALSE)

Use the same executor = "NOAA_SEDNA" with RiboPilot::new_project() for your own data. Keep projects on shared project or scratch space, not in your home folder. Disk space, caches, and offline use shows how to move Singularity’s caches and the reference folder there too.

On your own computer, open a tunnel to that node, using the name hostname printed:

ssh -N -L 8123:<node>:8123 <user>@sedna.nwfsc2.noaa.gov

Enter your password; if it works, nothing more is printed. Then open http://localhost:8123 in Chrome or Firefox. Leave both terminals open while you use the app.

The app has no login. Other SEDNA users can reach its port. Pick an unusual port number, and close the app when you are done.

Run the pipeline

Press Update. The run window shows a Slurm job script for the standard partition. Press Submit as job, or Save script to submit it yourself with sbatch. The job keeps going after you close the browser and the interactive session. Watch it with squeue, then press the reload button above the table to see new results. Submit a run as a job has the details.

The job needs Nextflow and Singularity. A job submitted from the app inherits the RiboPilot_deps environment. If you submit a saved script from a new terminal, run mamba activate RiboPilot_deps there first.