SeaWulf's Open OnDemand portal gives you 18 apps in your web browser: notebooks, R, VS Code, MATLAB and Mathematica, chemistry and structural biology tools, genome and data viewers, and a full Linux desktop. This page lists every app, what it's for, and the settings that matter when you launch it.
On this page: All apps at a glance · Launching an app · Coding and notebooks · Desktops and math · Chemistry and structural biology · Visualization and data · Troubleshooting
All apps at a glance
| App | What it's for | Menu group |
|---|---|---|
| Jupyter Notebook | Python notebooks | Servers |
| JupyterLab | The newer Jupyter interface, with files, notebooks, and terminals side by side | Sandbox |
| RStudio Server | R and RStudio in the browser | Servers |
| Code Server | VS Code in the browser | Servers |
| Galaxy | Your own private Galaxy server for point-and-click bioinformatics | Servers |
| SeaWulf Desktop | A full Linux desktop for any graphical program | Desktops |
| MATLAB | MATLAB's full desktop, R2016 through R2025a | GUIs |
| Mathematica | Wolfram Mathematica notebooks | Visualization |
| GNU Octave | Free, MATLAB-compatible numerical computing | Visualization |
| Gaussian/GaussView | Quantum chemistry calculations with a graphical builder | Chemistry |
| MOE | Molecular modeling and drug discovery | Molecular Modeling |
| UCSF ChimeraX | Molecular structure and cryo-EM map viewing and analysis | Visualization |
| PyMOL | Molecular visualization (open-source PyMOL) | Visualization |
| Coot | Building and fitting models into crystallography and cryo-EM maps | Visualization |
| Phenix | Crystallographic structure determination, with Coot | Visualization |
| ParaView | Scientific 3D visualization | Visualization |
| IGV | Integrative Genomics Viewer for sequencing data | Visualization |
| TensorBoard | Browse machine learning training logs | Visualization |
The portal also has the everyday tools: Seawulf Cluster Shell Access (a terminal), Files (with shortcuts to both your home folder and your scratch folder), Active Jobs, and Job Composer for batch jobs. See the OnDemand overview and Job Composer.
Launching an app
Every app runs as a job on SeaWulf's shared CPU queues, on either the 40-core Intel nodes or the 96-core AMD nodes. SeaWulf's OnDemand apps don't use GPUs; for GPU work, use NVwulf's OnDemand apps. Most launch forms ask for the same things:
| Setting | What to choose |
|---|---|
| Queue | short-40core-shared or short-96core-shared for up to 4 hours, long-…-shared for up to 24 hours, extended-…-shared for up to 3.5 days. Start with a short queue. |
| Hours | How long the session runs. It ends at this limit, so save your work before then. |
| Cores | 1 to 4 is plenty for most interactive work. Ask for more only if your program runs in parallel. |
| Memory | From 2 GB up to 164 GB. Start small; big requests wait longer in the queue. |
| Working directory | Where the app starts. Most default to your home folder, /gpfs/home/<netid>. Point it at your data. |
| Optional. Handy when the queue is busy and you want to know when the session starts. |
- Fill in the form and click Launch. The session card moves from Queued to Starting to Running.
- Click the connect button on the card to open the app in a new tab.
- When you're done, click Delete on the card so the cores go back to others. Closing the browser tab doesn't stop the job.
The full list of queues and limits is in the queue selection guide.
Coding and notebooks
| App | Good to know |
|---|---|
| Jupyter Notebook | The classic notebook interface. For your own Python packages, see Conda environments. More in Jupyter notebooks. |
| JupyterLab | The newer Jupyter interface. It's listed under Sandbox while it's being tested. The Modules to load box starts with jupyter/latest. |
| RStudio Server | Choose R 4.5.1 or 4.3.2. If an R package needs a system library to install, add its module in Additional modules (for example hdf5/1.12.1). More in RStudio. |
| Code Server | VS Code in the browser, with a terminal on the compute node. Set the working directory to your project folder. More in Code Server. |
| Galaxy | Starts your own private Galaxy server. Choose which SeaWulf folders Galaxy can see (home, scratch, projects). Galaxy keeps its data in ~/ondemand/data/galaxy unless you pick another folder; your home folder is only 20 GB, so point it at project space for real datasets. If you leave the password blank, one is created and saved in credentials.txt in that folder. Memory is typed as, for example, 16G. |
Desktops and math
| App | Good to know |
|---|---|
| SeaWulf Desktop | A full XFCE Linux desktop on a compute node, for any graphical program that doesn't have its own app. Open a terminal from the desktop to load modules and start programs. More in WolfieHPCDesktop. |
| MATLAB | Pick any version from R2016 to R2025a. The Modules box loads extra modules first, for example a compiler for MEX files. More in MATLAB. |
| Mathematica | Mathematica 13.1 with its full notebook interface, in a desktop session. |
| GNU Octave | Octave 9.3, a free alternative that runs most MATLAB code. |
Chemistry and structural biology
| App | Good to know |
|---|---|
| Gaussian/GaussView | Gaussian 16 (C.01) and GaussView 6 in a desktop. Build molecules in GaussView and run calculations from it. This is the one app that also offers whole-node queues (such as short-40core) for bigger calculations. Match the processor count in your Gaussian input to the cores you request. |
| MOE | Molecular Operating Environment 2024, for molecular modeling and drug discovery. Molecular dynamics and large databases can need a lot of memory. |
| UCSF ChimeraX | ChimeraX 1.11 for viewing and analyzing structures, sequences, and density maps. |
| PyMOL | Open-source PyMOL 3.1 for molecular graphics and figures. |
| Coot | Coot 1.1 for model building. Force software rendering is on by default; keep it on if you see EGL or OpenGL errors. |
| Phenix | Phenix 2.0 (CPU version) for crystallographic structure determination, with Coot loaded alongside it. |
Visualization and data
| App | Good to know |
|---|---|
| ParaView | ParaView 6.0.1 or 5.11.0 for visualizing simulation and imaging data, so large files stay on the cluster. |
| IGV | IGV 2.19 for viewing alignments, variants, and other genomic tracks. Set the working directory to where your BAM or VCF files are. |
| TensorBoard | Browse TensorFlow or PyTorch training logs. Set log directory to the folder with your events files. It runs on CPU only. |
/gpfs/scratch/<netid>) is handy for big data, but files older than 45 days are deleted. Keep anything you need in home or project space. See Storage layout & policies.Troubleshooting
My session stays Queued
The queue is busy. Try a short queue, fewer cores or less memory, or the other node type (40-core or 96-core).
A graphics app shows OpenGL or EGL errors, or a black window
These apps run without a GPU. For Coot, keep Force software rendering on. For other apps, delete the session and try again; if it keeps happening, open a ticket with the app name.
"Disk quota exceeded" or the app won't start
Your 20 GB home folder may be full; Galaxy data, conda environments, and caches add up fast. Check with myquota in a shell, then clean up or move data to project space. See Checking storage quotas.
I can't find my files
Check the working directory on the launch form. In the Files menu you can jump straight to your home or scratch folder.
My session ended in the middle of my work
Sessions stop at the hours you asked for. Save often, and use a long or extended queue for longer work.
Need an app that isn't here, or stuck on a problem? Open a ticket, or drop into HPC virtual office hours.
Related: OnDemand overview · OnDemand apps on NVwulf · OnDemand apps on ClinWulf · About SeaWulf