======================================================================================== GaugeFixer: Removing Unconstrained Degrees of Freedom in Sequence-Function Relationships ======================================================================================== ``GaugeFixer`` is a lightweight Python package designed to remove unconstrained degrees of freedom in the parameters of linear models for sequence-function relationships, where sequences are encoded using indicator functions on subsequences. History and Motivation ====================== Over the last few years, the Kinney and McCandlish labs have developed a theory of gauge freedoms and proposed a family of linear Gauges that includes commonly used Gauges for representing sequence-function relationships. - Anna Posfai, Juannan Zhou, David M. McCandlish, Justin B. Kinney (2025). Gauge fixing for sequence-function relationships. PLOS Computational Biology. `doi.org/10.1371 `_ - Anna Posfai, David M. McCandlish, Justin B. Kinney (2025). Symmetry, gauge freedoms, and the interpretability of sequence-function relationships. Physical Review Research. `doi.org/10.1103 `_ While the theory is well established, computational tools for fixing the gauge for a subset of linear gauges are only partially present in `MAVE-NN `_, a software package for quantitative modeling of sequence-function relationships. ``GaugeFixer`` was born out of the need for a comprehensive, efficient, and user-friendly set of tools for fixing the gauge in a way that is independent of the specific implementation of the model. Installation ============ We recommend using a new independent environment with `Python >3.10`, as used during development and testing of ``GaugeFixer``, to minimize problems with dependencies. For instance, one can create and activate a new conda environment as follows: :: $ conda create -n gauge python=3.10 $ conda activate gauge ``GaugeFixer`` is available on PyPI and installable through the ``pip`` package manager: :: $ pip install gaugefixer You can also install the latest or specific versions from `GitHub `_ as follows: :: $ git clone https://github.com/jbkinney/gaugefixer.git and install it in the current Python environment: :: $ cd gaugefixer $ pip install . For developers, tests can be run using ``pytest``: :: $ pytest test Citation ======== - Carlos Martí-Gómez, David M. McCandlish, Justin B. Kinney (2025). GaugeFixer: Removing unconstrained degrees of freedom in sequence-function relationships. bioRxiv. `doi.org/10.64898/2025.12.08.693054 `_ .. toctree:: :maxdepth: 1 :caption: Table of Contents usage/tutorial.ipynb usage/benchmarking.ipynb usage/example.ipynb api