EPyR Tools: Electron Paramagnetic Resonance (EPR) Tools in Python

    

License	Tests	Documentation	Version

What is EPyR Tools?

EPyR Tools is a Python package for analyzing Electron Paramagnetic Resonance (EPR) spectroscopy data from Bruker spectrometers. It covers the full path from raw instrument files to publication-ready results: loading BES3T and ESP/WinEPR binary formats, baseline correction, lineshape and T1/T2 relaxation fitting, FFT-based analysis of pulse-EPR time-domain data, and FAIR-compliant export to CSV, JSON, and HDF5.

The package targets EPR researchers who want a reproducible, scriptable Python workflow in place of proprietary vendor software, with a command-line interface for routine processing and a documented Python API for custom analysis.

Key Features

Data Loading & FAIR Conversion

• Read Bruker BES3T (.DTA/.DSC) and ESP/WinEPR (.spc/.par) files, 1D and 2D, with automatic format detection
• epyr.fair: export to CSV, JSON, and HDF5 with standardized, complete metadata
• Batch conversion of entire directories
• Plugin architecture for adding new file formats

Baseline Correction (epyr.baseline)

• Polynomial, exponential, bi-exponential, and stretched-exponential models, for 1D and 2D data
• Automatic model selection ranked by AIC/BIC/R²
• Manual and interactive baseline-region selection

Lineshape Analysis & Fitting (epyr.lineshapes)

• Gaussian, Lorentzian, true Voigt (Faddeeva-function convolution), and pseudo-Voigt lineshapes
• Absorption and 1st/2nd derivative forms, phase mixing, optional affine baseline term
• fit_epr_signal() and fit_multiple_shapes() for single-model and model-comparison fitting

T1/T2 Relaxation Fitting (epyr.relaxation, new in v0.4.0)

• Six decay/recovery models: mono- and stretched-exponential, bi-exponential, inversion/saturation recovery, and combined homogeneous/spectral-diffusion (Gamma0/GammaG) echo decay
• fit_relaxation() for a single model, fit_multiple_decays() to rank candidates by reduced chi-squared
• Fit results print a readable parameter summary directly

Signal Processing (epyr.signalprocessing)

• FFT-based frequency analysis for pulse-EPR time-domain data: Rabi oscillations, DEER, HYSCORE
• 1D and 2D FFT modes, apodization windows (Hann, Hamming, Blackman, Kaiser), automatic time-unit detection, zero-padding
• Power spectral density (Welch, periodogram) and spectrogram analysis

Physics & Units (epyr.physics)

• CODATA 2022 physical constants in SI and CGS units (GFREE, BMAGN, NMAGN, PLANCK, ...)
• Field/frequency conversion (mT ↔ MHz ↔ cm⁻¹) via unitconvert() and dedicated helpers

Command-Line Interface

• Nine commands covering the full workflow: epyr-convert, epyr-baseline, epyr-batch-convert, epyr-config, epyr-info, epyr-isotopes, epyr-plot, epyr-validate, epyrview
• epyr-plot --interactive --measure: click-to-measure delta x/y distance tool

Performance & Isotope Database

• OptimizedLoader / DataCache for large files, with memory monitoring and streaming
• epyr.isotopes / epyr-isotopes: interactive periodic-table GUI with NMR frequency calculator and X/Q/W-band presets

What's New in v0.4.0

Version 0.4.0 adds the epyr.relaxation package for T1/T2 relaxation fitting, complementing the existing field-domain lineshape fitting in epyr.lineshapes.fitting:

from epyr.relaxation import fit_relaxation, fit_multiple_decays

# Single model
result = fit_relaxation(t, y, model="stretched_exponential")
print(result)
# === Relaxation Fit Results - stretched_exponential ===
# Success: True
# R2 = 0.998452
# ...

# Compare candidate models, ranked by reduced chi-squared (not R-squared,
# which is biased toward models with more free parameters)
results = fit_multiple_decays(t, y)
print(results)
# model                  success  R2        chi2       amplitude  T      ...
# mono_exponential       True     0.998391  0.0004842  2.005      1.310  ...
# stretched_exponential  True     0.998452  0.000476   1.976      1.309  ...

Fit plots in both epyr.relaxation and epyr.lineshapes.fitting now follow matplotlib.rcParams for figure size, marker size, line width, and font size, instead of a fixed layout.

See docs/release_notes/v0.4.0.rst for full details, or docs/release_notes.rst for the complete version history.

Installation

Prerequisites

• Python 3.8 or higher
• NumPy, SciPy, matplotlib, pandas, h5py (installed automatically)

Quick Install

pip install epyr-tools

Development Installation

git clone https://github.com/BertainaS/epyrtools.git
cd epyrtools

# Install with development dependencies
pip install -e ".[dev,docs]"

# Set up pre-commit hooks
pre-commit install

Verification

epyr --help
epyr-info
make test

Getting Started

1. Loading Data

import epyr

# Open a file dialog to select a .dta, .dsc, .spc, or .par file
x, y, params, filepath = epyr.eprload()

# Or specify a path directly:
# x, y, params, filepath = epyr.eprload('path/to/data.dsc')

2. Converting to FAIR Formats

from epyr.fair import convert_bruker_to_fair

convert_bruker_to_fair('path/to/data.dsc', output_dir='path/to/output')

3. Baseline Correction

import epyr

x, y, params, filepath = epyr.eprload("data.dsc")

# Automatic model selection
corrected, baseline, model_info = epyr.baseline.baseline_auto_1d(x, y, params)

# Or a specific model with manual region exclusion (e.g. signal regions, in mT)
corrected, baseline = epyr.baseline.baseline_polynomial_1d(
    x, y, params,
    manual_regions=[(3340, 3360), (3380, 3400)],
    region_mode='exclude',
    order=2,
)

4. Lineshape Fitting

from epyr.lineshapes import fit_epr_signal, fit_multiple_shapes

x, y, params, filepath = epyr.eprload('data.DTA')

# Single model
result = fit_epr_signal(x, y, 'gaussian')
print(result.summary())

# 1st-derivative signal with adjustable phase, comparing all lineshapes
results = fit_multiple_shapes(x, y, derivative=1, fit_phase=True)

5. T1/T2 Relaxation Fitting

from epyr.relaxation import fit_relaxation, fit_multiple_decays

t, y, params, filepath = epyr.eprload('echo_decay.DTA')
y = abs(y)  # take the magnitude of a complex echo signal

result = fit_relaxation(t, y, model="stretched_exponential")
results = fit_multiple_decays(t, y)  # compare mono/stretched/bi-exponential

6. Time-Domain Signal Processing (FFT)

from epyr.signalprocessing import analyze_frequencies

t, y, params, filepath = epyr.eprload('rabi_oscillation.DTA')
result = analyze_frequencies(t, y, window='hann', zero_padding=4)
print(f"Dominant frequency: {result['dominant_frequencies'][0]:.3f} MHz")

7. Plotting and the CLI

import epyr

x, y, params, filepath = epyr.eprload("data.dsc")
epyr.plot_1d(x, y, params, title="EPR Spectrum")

# Interactive plot with click-to-measure delta x/y
epyr-plot spectrum.dsc --interactive --measure

# Batch FAIR conversion
epyr-batch-convert ./data --formats csv,json,hdf5

Tutorials & Examples

Jupyter Notebook Series

An eight-notebook tutorial series in examples/notebooks/, using real experimental data from examples/data/. Notebooks are committed without outputs; run them to generate figures.

cd examples/notebooks
jupyter lab 00_Tutorial_Series_Index.ipynb  # index and navigation

Notebook	Topic
01_Loading_and_Plotting.ipynb	eprload, parameter inspection, 1D/2D plotting
02_Baseline_Correction.ipynb	Polynomial, automatic, and exponential baselines
03_Lineshape_Analysis_and_Fitting.ipynb	Gaussian/Lorentzian/Voigt, derivatives, fitting
04_Relaxation_Fitting.ipynb	T1/T2 decay/recovery models (new in v0.4.0)
05_Signal_Processing_and_FFT.ipynb	Frequency analysis of Rabi data, apodization
06_FAIR_Conversion_and_Export.ipynb	CSV/JSON/HDF5 export and validation
07_Physics_Units_and_Constants.ipynb	CODATA constants, field/frequency conversions

Standalone Example Scripts

Six short, self-contained scripts in examples/clean/ exercising the public API end to end:

python examples/clean/01_basic_loading_and_plotting.py
python examples/clean/02_baseline_and_fitting.py
python examples/clean/03_advanced_fft_windows.py
python examples/clean/04_interactive_2d_slicer.py
python examples/clean/05_rabi_frequency_analysis.py
python examples/clean/06_relaxation_fitting.py

See docs/tutorials/clean_examples.rst for a description of each script.

Project Structure

epyrtools/
├── epyr/                          # Main package
│   ├── eprload.py                 # Core data loading (BES3T, ESP formats)
│   ├── eprplot.py                 # EPR plotting (1D, 2D map, waterfall, slicer)
│   ├── cli.py                     # Command-line interface (9 commands)
│   ├── config.py                  # Hierarchical configuration system
│   ├── performance.py             # OptimizedLoader, DataCache
│   ├── plugins.py                 # Plugin architecture
│   ├── logging_config.py          # Centralized logging
│   ├── isotope_gui.py             # Interactive isotope database GUI
│   ├── baseline/                  # Baseline correction (correction, selection, models, interactive)
│   ├── lineshapes/                # Gaussian, Lorentzian, Voigt, pseudo-Voigt, fitting
│   ├── relaxation/                # T1/T2 decay/recovery models and fitting
│   ├── signalprocessing/          # FFT frequency analysis, apodization windows
│   ├── physics/                   # CODATA constants and unit conversions
│   ├── fair/                      # FAIR conversion, exporters, validation
│   └── sub/                       # Bruker BES3T/ESP format loaders
├── docs/                          # Sphinx documentation and tutorials
├── examples/
│   ├── notebooks/                 # Jupyter tutorial series
│   ├── clean/                     # Six standalone end-to-end scripts
│   └── data/                      # Real EPR measurement files (CW, pulse, 2D)
├── tests/                         # Test suite (369 tests; smoke/standard/deep/scientific)
└── pyproject.toml                 # Packaging, dependencies, entry points

Documentation

• Full documentation: guides and API reference (Sphinx)
• User guide: workflows and step-by-step tutorials
• CLI reference: command-line interface
• API reference: public API
• Release notes: version history

Testing & Quality

EPyR Tools follows a 4-level testing protocol (pytest -m smoke|standard|deep|scientific), with 369 tests covering basic functionality, broad feature coverage, edge cases, and scientific validation against NIST/CODATA values.

make test        # full suite
make test-cov    # with coverage report
make quality     # lint, type-check, security

Contributing & Support

• Issues: GitHub Issues
• Contributing guide: docs/contributing.rst

License

This project is licensed under the MIT License, see LICENSE for details.

Contributors

Lead Developer & Maintainer:

• Sylvain Bertaina, sylvain.bertaina@cnrs.fr

Affiliation:

• Magnetism Group (MAG), IM2NP Laboratory
• CNRS (Centre National de la Recherche Scientifique)

---

EPyR Tools: EPR data analysis in Python.