Capture, detect and quantify gel electrophoresis images
Work for basic cameras; not yet gel docs
Features:
- Supports DNA, RNA and protein gels.
- Camera snapshots, including multi-exposure HDR for dynamic range
- Detection of bands using ML, and gaussian mixture distribution to figure out if the are at an angle. The gel is modelled by NURBS, fitting band angles and ladder band positions
- Quantification of densitys and molarities, taking gel warping into account
- Quick compute of relative mass and molarity ratios
File formats:
| Format | Read | Write |
|---|---|---|
.gel.zip — OpenGel's own container (images, analysis, metadata) |
✅ | ✅ |
.scn / .mscn — Bio-Rad Image Lab scan, single- and multi-channel |
✅ | — |
.sscn / .smscn — Image Lab "secured" scan (same container, signed) |
✅ | — |
| PNG, JPEG, TIFF, BMP and friends — a loose image, imported as one gel | ✅ | — |
| PDF — report of the current analysis | — | ✅ |
Hardware support:
- All auto-discoverable cameras through nu-manager,
- Plain USB webcams (UVC)
- Bio-Rad Gel Doc EZ (specialized view)
- If you want support for another camera, make a github issue on the nu-manager repository
Gel doc requires additional rules on Linux:
gel udev-rules | sudo tee /etc/udev/rules.d/60-opengel.rules
sudo udevadm control --reload-rules && sudo udevadm trigger
The pretrained band-detection model (assets/models/*.bpk) is stored with
Git LFS, so you must install it before cloning (or
run git lfs pull after) — otherwise you only get a small pointer file and the
build/run fails to load the model.
# Install Git LFS, then fetch the model weights:
brew install git-lfs # macOS (Linux: sudo apt-get install git-lfs)
git lfs install # once per machine
git lfs pull # download the model into an already-cloned repocargo build --release # optimized build (USB camera backend on by default)
# GUI (the default binary)
cargo run --release
# CLI
cargo run --release --bin gelThe crate ships two binaries — opengel (the desktop GUI) and gel (the CLI).
default-run points bare cargo run at the GUI; pass --bin gel for the CLI.
The USB camera backends are on by default on every platform. The webcam backend
needs libv4l-dev at build time on Linux (sudo apt-get install libv4l-dev);
build with --no-default-features to drop it for a headless build without the
v4l toolchain. nu-manager's cameras need no build-time system deps, but on Linux
they do need raw USB access. The Debian package installs the udev rules for you;
for a raw binary install, generate and install them yourself:
cargo run --release --bin gel -- udev-rules \
| sudo tee /etc/udev/rules.d/60-opengel-numanager.rules
sudo udevadm control --reload-rules && sudo udevadm trigger --subsystem-match=usb
# then replug the cameraThe rules are generated, not checked in: gel udev-rules derives them from
nu-manager's own list of claimed USB vendor ids — the same declaration that
decides which drivers probe the bus — so they cannot fall behind as nu-manager
gains device support. Regenerate after updating nu-manager.
The initial detection of bands is done using the pretrained ML model of GelGenie, converted to Rust and adapted for the NURBS model. It is an important component and it would thus be fair you could cite:
Aquilina, M., Wu, N. J. W., Kwan, K., Bušić, F., Dodd, J., Nicolás-Sáenz, L., O'Callaghan, A., Bankhead, P., & Dunn, K. E. (2025). GelGenie: an AI-powered framework for gel electrophoresis image analysis. Nature Communications, 16, 4087. https://doi.org/10.1038/s41467-025-59189-0
In addition, please cite this git repository. So you could write something like: Gels were analyzed using https://github.com/henriksson-lab/opengel, using GelGenie[1] for band detection
- Code is by default under MIT license
- Code under src/gelgenie is a Rust conversion of GelGenie, which is under Apache-2.0 license.
Note that code has been produced using agentic AI; in case you wish to copy out any part of the code, please first please review the code for accidental reuse of copyrighted material.