This repository hosts code for a number of different projects related to minimizers.

1. Python code in py/header.py to evaluate the density lower bound of

   Bryce Kille, Ragnar Groot Koerkamp, Drake McAdams, Alan Liu, and Todd J Treangen. A near-tight Lower Bound on the Density of Forward Sampling Schemes. Bioinformatics, December 2024. https://doi.org/10.1093/bioinformatics/btae736

2. In src/lib.rs, reference implementations for the schemes presented in the mod-minimizer paper with Giulio Ermanno Pibiri.
   • See also this blogpost.
   • These are also used for the lower bound paper with Bryce Kille:
   • Python bindings available in src/py.rs.

   Ragnar Groot Koerkamp, and Giulio Ermanno Pibiri. The Mod-Minimizer: A Simple and Efficient Sampling Algorithm for Long k-Mers. WABI 2024. https://doi.org/10.4230/LIPIcs.WABI.2024.11

   Ragnar Groot Koerkamp, Daniel Liu, and Giulio Ermanno Pibiri. The Open-Closed Mod-Minimizer Algorithm. Algorithms for Molecular Biology, March 2025. https://doi.org/10.1093/bioinformatics/btae736

3. An implementation (src/schemes/anchors.rs) and experiments (py/sus-anchors-1.py) of SUS-anchors (arxiv).

   Ragnar Groot Koerkamp. The Anti-Lexicographic SUS-anchor: A near-Optimal k=1 Sampling Scheme. Arxiv 2026. https://doi.org/10.48550/ARXIV.2606.01190

4. Ongoing research on new sampling schemes.
   • Plots and experiments are in py/.

The SIMD-based random-minimizers implementation corresponding to this post has moved to https://github.com/rust-seq/simd-minimizers.

Implemented minimizer schemes

• Random minimizers.
• two versions of asymptotically optimal Rotational minimizers (Marçais et al., 2018) .
• Miniception, and a small slightly improved variant of it.
• Decycling and double decycling based minimizers (Pellow et al., 2023).
• Bidirectional anchors (Loukides et al., 2023)
• Mod-sampling, with lr-minimizers and mod-minimizers (our work).
• SUS-anchors

Reproducing SUS-anchor results

1. Install maturin, the build system we use for creating python libraries out of Rust code.
2. Create a python environment: just py-init (python -m venv -env.) and load it source .env/bin/activate
3. Run just py (maturin develop -r) to build the python library.
4. cd py and run ./sus-anchors-1.py to reproduce the sus-anchors-1.svg plot in the SUS-anchors paper. Lower n to 100000 or so for faster results.