rxnfit 0.2.0

set up theoretical reaction kinetics equations

Overviews

"rxnfit" is a Python module that builds ordinary differential equations (ODEs) for reaction kinetics from elementary reaction formulas and rate constants given in CSV format, and computes concentration time courses numerically.

• When all rate constants are known, the time evolution of each species’ concentration is obtained by numerical integration of the ODEs (e.g. with scipy.integrate.solve_ivp).
• When some rate constants are unknown, they can be estimated by fitting the ODE solution to experimental concentration time-course data (e.g. with scipy.optimize.minimize).
• A rate constant may be given as an expression in other rate constants (e.g. k₂ = 2·k₁), or as a function of time t via an override CSV.

Documentation : https://pejpohno.github.io/rxnfit/
Tutorial : in preparation
How to cite : Ohno, M. rxnfit. GitHub. https://github.com/PEJpOhno/rxnfit (2026).

Current version and requirements

• current version = 0.2.0
• pyhon >=3.12

[dependencies]

• NumPy
• pandas
• SciPy
• SymPy
• Matplotlib
• Optuna

Copyright and license

Copyright (c) 2025 Mitsuru Ohno Released under the BSD-3 license, license that can be found in the LICENSE file.

Installation

Create and activate a Python virtual environment, then

pip install rxnfit

If you want to give it a quick try, just copy or move the example directory from the cloned repo to wherever it's convenient. Then, activate your virtual environment, start Jupyter Notebook, and open one of the sample scripts inside "examples" directry to test it out.

Usage

Prepare csv file

1. Describe reaction formula in csv format. THe first row shoud be header. The first column and the second column should be reaction ID (named RID) and the rate constant (named k). Other column names can be omitted.
2. The reactants are putted in alternately with the coeficient and the chemical species. Each reaction should be described in one row.
3. To separate the reactants and the products, two ">" are setted with two columns. As with reaction SMARTS, you may write reaction conditions between the two ">"; in the current version of rxnfit these are unused.
4. The products are putted in alternately with the coeficient and the chemical species. Note the reaction formula should be inputted by the left filling. To describe chemical species, it is better to avoid to use arithmetic symbols such as "+", "-" and "*" (e.g. use "a"(nion) in place of "-"). The coefficient '1' can be substituted for blank.
5. Each elementary reaction’s rate constant must be specified in the format ‘k + RID’. If left empty, the system will automatically assign ‘k + RID’.

The rate constant k_m can also be defined as a function of another elementary reaction rate constant k_n, or as a function of time t.
The example of a csv format is as follows. Attached "examples/sample_data/sample_rxn_ref1.csv" and "examples/sample_data/refX/sample_rxn_refX.csv" are also available for demonstration.

example of a csv format

RID, k,,,,,,,,,,
1,0.054,,AcOEt,,OHa1,>,>,,AcOa1,,EtOH
2,1.4,,AcOiPr,,OHa1,>,>,,AcOa1,,iPrOH
3,0.031,,EGOAc2,2,OHa1,>,>,2,AcOa1,,EG

Modules and classes

The following items are those prominently used in the example notebooks.

rxn_reader

Reads reaction CSV and builds symbolic ODE system.

item	name	description
class	RxnToODE	Builds differential-type rate equations (strings) from reaction CSV, then ODE system expressions (SymPy). Attributes: rate_consts_dict, sympy_symbol_dict, sys_odes_dict, function_names.
function	get_reactions	Parse reaction CSV into list of reaction data (RID, k, reactants, products, conditions).
function	get_unique_species	Extract unique species names from reaction list.
function	rate_constants	Extract rate constant key and value (float or expression string) per reaction.

build_ode

Builds numerical ODE RHS from symbolic system; supports rate-constant overrides and time-dependent k(t).

item	name	description
class	RxnODEbuild	Extends RxnToODE. Builds callable ODE RHS for scipy.integrate.solve_ivp. Optional rate_const_overrides (dict or CSV path) and rate_const_overrides_encoding. Methods: create_ode_system(), create_ode_system_with_rate_consts(), get_ode_system(), get_ode_info().
function	create_system_rhs	Build (t, y) RHS for solve_ivp from ODE functions dict. Optional rate_const_values (dict or callable(t)) and symbolic_rate_const_keys for variable or time-dependent rate constants.

expdata_reader

Loads and aligns experimental time-course data.

item	name	description
function	expdata_read	Read list of DataFrames into list of (t_list, C_exp_list) per dataset.
function	get_y0_from_expdata	Initial concentrations per dataset in function_names order.

expdata_fit_sci

Fits symbolic rate constants to experimental data (scipy.optimize.minimize).

item	name	description
class	ExpDataFitSci	Multi-dataset fitting. run_fit(p0, ...) returns (result, param_info, fit_metrics). result has .fun (RSS), .tss, .r2; fit_metrics is dict with keys 'rss', 'tss', 'r2'. get_solver_config_args(dataset_index), get_fitted_rate_const_dict(result). plot_fitted_solution(expdata_df, ...) plots per-dataset y0 curves after run_fit. to_dataframe_list() returns list of DataFrames (one per dataset). When model has k(t), get_solver_config_args() adds rate_const_values (callable) and symbolic_rate_const_keys after run_fit().

solv_ode

Numerical integration and plotting of ODE solutions.

item	name	description
class	SolverConfig	Dataclass: y0, t_span, t_eval, method, rtol. Optional: rate_const_values (dict or callable(t)), symbolic_rate_const_keys (both or neither).
class	RxnODEsolver	Integrates ODE with builder and config. solve_system(), to_dataframe_list() returning list of DataFrames, eval_fit_metrics(expdata_df, ...) returning dict with 'rss', 'tss', 'r2', solution_plot(). When config has rate_const_values and symbolic_rate_const_keys, uses rate-constants ODE path.

p0_opt_fit

Optimizes initial parameter values (p0) for rate constants using Optuna, then fits with ExpDataFitSci.

item	name	description
class	P0OptFit	Optimizes p0 via Optuna (e.g. suggest_float with log=True) and runs ExpDataFitSci.run_fit with the best p0. optimize(n_trials, show_progress_bar, ...) returns (dict of variable → (initial, fitted), fit_metrics). fit_metrics is a dict with keys 'rss', 'tss', 'r2'. optuna_log() returns per-trial log.

References

• Python: Python Software Foundation. (2024). Python (Version 3.13) [Computer software]. https://www.python.org/
• NumPy: Harris, C. R., Millman, K. J., van der Walt, S. J., et al. (2020). Array programming with NumPy. Nature, 585(7825), 357–362. https://doi.org/10.1038/s41586-020-2649-2
• pandas: McKinney, W. (2010). Data Structures for Statistical Computing in Python. Proceedings of the 9th Python in Science Conference, 56–61. https://doi.org/10.25080/Majora-92bf1922-00a
• SciPy: Virtanen, P., Gommers, R., Oliphant, T. E., et al. (2020). SciPy 1.0: Fundamental Algorithms for Scientific Computing in Python. Nature Methods, 17(3), 261–272. https://doi.org/10.1038/s41592-019-0686-2
• SymPy: Meurer, A., Smith, C. P., Paprocki, M., et al. (2017). SymPy: symbolic computing in Python. PeerJ Computer Science, 3, e103. https://doi.org/10.7717/peerj-cs.103
• Matplotlib: Hunter, J. D. (2007). Matplotlib: A 2D graphics environment. Computing in Science & Engineering, 9(3), 90–95. https://doi.org/10.1109/MCSE.2007.55
• Optuna: Akiba, T., Sano, S., Yanase, T., Ohta, T., & Koyama, M. (2019). Optuna: A next-generation hyperparameter optimization framework. Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining (KDD '19), 2623–2631. https://doi.org/10.1145/3292500.3330701

Acknowledgements

This module and its accompanying documentation were developed with the support of Cursor’s AI-assisted tools.