enzyme-fit

Fit Michaelis-Menten and substrate-inhibition (Haldane) enzyme kinetics models to initial-velocity data, with objective model selection (AIC), a built-in Lineweaver-Burk comparison, and a publication-ready report.

Problem

Determining Km and Vmax from initial-velocity vs. substrate-concentration data is one of the most common quantitative tasks in enzymology and food biochemistry (fermentation, enzymatic browning, hydrolytic processing enzymes). It is still routinely done via the Lineweaver-Burk double-reciprocal linearization (1/v vs 1/[S]) — a method known since the 1970s to give biased parameter estimates because it disproportionately weights low-velocity (high 1/v, high-noise) points, compared to direct nonlinear regression on the untransformed data. Substrate inhibition at high concentrations (common for many enzymes) also goes undetected unless a dedicated model is fit and compared objectively. enzyme-fit automates nonlinear regression for Michaelis-Menten and Haldane substrate-inhibition models, selects the best-describing model by AIC, and reports a Lineweaver-Burk panel side-by-side with the nonlinear fit purely as a diagnostic — making the difference, and the recommended method, explicit.

Installation

pip install enzyme-fit

Or from source:

git clone https://github.com/karenkhachatryan-lab/enzyme-fit.git
cd enzyme-fit
pip install -e ".[dev]"

Usage

Input CSV with columns substrate, velocity, and optionally velocity_std (any consistent concentration/rate units):

substrate,velocity,velocity_std
2,577.6,10.2
4,862.4,12.1
6,1032.0,11.8
8,1144.5,13.4
10,1224.6,14.0

Fit models and generate a report:

enzyme-fit fit data.csv --output report

This produces:

• report.pdf — velocity vs. substrate curve with all fitted models, a Lineweaver-Burk diagnostic panel, and a residuals plot for the best model,
• report.json — fitted parameters, metrics (R², RMSE, AIC), kinetics classification, and the Lineweaver-Burk comparison values.

Print citation information:

enzyme-fit cite

Desktop GUI (optional)

pip install "enzyme-fit[gui]"
enzyme-fit gui

Opens a desktop window (CustomTkinter) to load a CSV, pick models, view the kinetics plot and metrics live, and save the PDF/PNG/JSON outputs — no command-line arguments needed.

For users without Python, a prebuilt standalone Windows GUI is attached as a .zip to each GitHub Release — download, extract, and run enzyme-fit-gui.exe, no installation needed. To build it yourself instead, see packaging/build_exe.ps1 (PyInstaller).

Models

• Michaelis-Menten: v = Vmax·[S] / (Km + [S]) — 2 parameters, the standard model for the vast majority of enzymes.
• Haldane (substrate inhibition): v = Vmax·[S] / (Km + [S] + [S]²/Ki) — 3 parameters; velocity rises then declines at high substrate concentration, common for many hydrolases and oxidases at supra-physiological substrate levels.

Both models are fit on the full dataset, so their AIC values are always directly comparable — no restricted-range caveat (unlike isotherm-fit's BET model).

Project status

Early development (v0.1.0) — MVP scope: 2 models, CSV loader, AIC-based model selection, kinetics classification (inhibition detection + optimal substrate concentration), Lineweaver-Burk diagnostic panel, PDF/PNG + JSON report generation, CLI via Typer, optional CustomTkinter desktop GUI. See CHANGELOG.md.

Citing this software

If you use enzyme-fit in your research, please cite it — see CITATION.cff or run enzyme-fit cite for the formatted citation and BibTeX entry. DOI: 10.5281/zenodo.21724603.

Contributing and support

Bug reports, feature requests, and usage questions are all welcome via GitHub Issues.

License

MIT — see LICENSE.