synth-nmr

NMR spectroscopy calculations for protein structures

A lightweight, standalone Python package for calculating NMR observables from protein structures. Originally extracted from the synth-pdb package to provide a focused toolkit that works with any protein structure source.

Read the full documentation here!

Features

• NOE Calculations: Synthetic NOE distance restraints
• Relaxation Rates: R1, R2, and heteronuclear NOE predictions
• Chemical Shifts: Integrates to SHIFTX2 predictions with SPARTA+ empirical fallback
• J-Couplings: Scalar coupling prediction using the Karplus equation with Vuister & Bax (1993) parameters.
• RDC Calculations: Prediction of residual dipolar couplings
• MD Trajectory / Ensemble NMR: S², ensemble-averaged shifts, NOEs (r⁻⁶), and RDCs from NMR ensembles or MD trajectories
• NEF I/O: Read and write NMR Exchange Format files
• Secondary Structure: Automatic classification for enhanced predictions

Interactive Tutorials

Try out synth-nmr immediately in your browser using Google Colab!

• Basic NMR Prediction
• Advanced Observables: J-Couplings, NOEs, and RDCs
• Relaxation & Dynamics Analysis
• Ensemble NMR Analysis: Proteins in Motion

AI Assistant & Interactive Documentation

NotebookLM Source Guide
We've assembled the repository documentation, roadmap, and core physics code into a single curated text file: synth_nmr_source_guide.txt. You can upload this file directly to Google's NotebookLM to instantly create an interactive, AI-powered study guide. Use it to chat with the codebase, learn the underlying NMR spectroscopy physics, or ask how to use the CLI and API!

Installation

pip install synth-nmr

For improved performance with JIT compilation:

pip install synth-nmr[performance]

For MD trajectory / ensemble NMR analysis:

pip install synth-nmr[trajectory]

Command-Line Interface

synth-nmr provides a command-line interface for common tasks, allowing you to perform calculations directly from your terminal.

Usage

You can run synth-nmr CLI commands directly or enter an interactive mode.

Non-Interactive Mode

Execute commands by passing them as arguments to the synth_nmr.synth_nmr_cli module:

python -m synth_nmr.synth_nmr_cli <command> [arguments]

Examples:

1. Read a PDB file and calculate RDCs:

   python -m synth_nmr.synth_nmr_cli read pdb protein.pdb calculate rdc 10.0 0.5

2. Read a PDB file and predict chemical shifts:

   python -m synth_nmr.synth_nmr_cli read pdb protein.pdb predict shifts

3. Read a PDB file and calculate J-couplings:

   python -m synth_nmr.synth_nmr_cli read pdb protein.pdb calculate j-coupling

Interactive Mode

To enter interactive mode, run the CLI without any arguments:

python -m synth_nmr.synth_nmr_cli

Once in interactive mode, you will see a SynthNMR> prompt. Type help to see available commands:

SynthNMR> help
Commands:
  read pdb <filename>
  calculate rdc [Da] [R]
  predict shifts
  calculate j-coupling
  exit
SynthNMR> read pdb protein.pdb
SynthNMR> calculate rdc 10.0 0.5
SynthNMR> exit

Available Commands

• read pdb <filename>: Loads a protein structure from the specified PDB file. This command must be executed before any calculation commands.
• calculate rdc [Da] [R]: Calculates Residual Dipolar Couplings.
  • Da: (Optional) Axial component of the alignment tensor in Hz (default: 10.0).
  • R: (Optional) Rhombicity of the alignment tensor (dimensionless) (default: 0.5).
• predict shifts: Predicts chemical shifts using SPARTA+ with ring current corrections.
• calculate j-coupling: Calculates ³J(HN-Hα) couplings using the Karplus equation.
• help: (Interactive mode only) Displays a list of available commands.
• exit: (Interactive mode only) Exits the CLI.

Quick Start

import biotite.structure.io as strucio
from synth_nmr import (
    calculate_synthetic_noes,
    calculate_relaxation_rates,
    predict_chemical_shifts,
    calculate_hn_ha_coupling,
    calculate_rdcs
)

# Load a protein structure
structure = strucio.load_structure("protein.pdb")

# Calculate NOEs
noes = calculate_synthetic_noes(structure, cutoff=5.0)

# Predict relaxation rates
relaxation = calculate_relaxation_rates(
    structure,
    field_strength=600.0,  # MHz
    temperature=298.0,      # K
    correlation_time=5.0    # ns
)

# Predict chemical shifts
shifts = predict_chemical_shifts(structure)

# Calculate J-couplings
j_couplings = calculate_hn_ha_coupling(structure)

# Predict RDCs
rdcs = calculate_rdcs(
    structure,
    Da=10.0, # Axial component of alignment tensor (Hz)
    R=0.5    # Rhombic component of alignment tensor
)

Requirements

• Python ≥ 3.8
• NumPy ≥ 1.20
• Biotite ≥ 0.35.0
• Numba ≥ 0.55.0 (optional, for performance)

Documentation

Core Functions

calculate_synthetic_noes(structure, cutoff=5.0)

Calculate synthetic NOE distance restraints.

Parameters:

• structure: biotite AtomArray
• cutoff: Distance cutoff in Ångströms (default: 5.0)

Returns: Dictionary of NOE restraints

calculate_relaxation_rates(structure, field_strength, temperature, correlation_time)

Predict NMR relaxation rates (R1, R2, heteronuclear NOE).

Parameters:

• structure: biotite AtomArray
• field_strength: Spectrometer frequency in MHz
• temperature: Temperature in Kelvin
• correlation_time: Molecular correlation time in nanoseconds

Returns: Dictionary of relaxation rates per residue

predict_chemical_shifts(structure)

Predict chemical shifts using SPARTA+ with ring current corrections.

Parameters:

• structure: biotite AtomArray

Returns: Dictionary of chemical shifts by residue and atom type

calculate_hn_ha_coupling(structure)

Calculate ³J(HN-Hα) couplings using the Karplus equation.

Parameters:

• structure: biotite AtomArray

Returns: Dictionary of J-coupling values per residue

calculate_rdcs(structure, Da, R)

Predict residual dipolar couplings (RDCs) for backbone N-H vectors.

Parameters:

• structure: biotite AtomArray
• Da: Axial component of the alignment tensor in Hz
• R: Rhombicity of the alignment tensor (dimensionless)

Returns: Dictionary of RDC values per residue

Use Cases

• Structure Validation: Compare predicted vs experimental NMR data
• MD Analysis: Calculate NMR observables from molecular dynamics trajectories
• Protein Design: Predict NMR properties of designed structures
• Data Integration: Generate synthetic NMR data for machine learning

Compatibility

Works with protein structures from any source:

• PDB files
• AlphaFold predictions
• Molecular dynamics simulations
• De novo structure generation (e.g., synth-pdb)

Citation

If you use synth-nmr in your research, please cite:

@software{synth_nmr,
  author = {Elkins, George},
  title = {synth-nmr: NMR spectroscopy calculations for protein structures},
  year = {2026},
  doi = {10.5281/zenodo.18855129},
  url = {https://github.com/elkins/synth-nmr}
}

License

MIT License - see LICENSE file for details

Related Projects

• synth-pdb - Synthetic protein structure generation
• Biotite - Computational biology toolkit

References

This package relies on the following peer-reviewed research:

• SPARTA+: For chemical shift predictions.

  Yang, Y., & Bax, A. (2011). Journal of Biomolecular NMR, 51(3), 259–274.

• Karplus Equation: For J-coupling calculations.

  Karplus, M. (1959). The Journal of Chemical Physics, 30(1), 11–15.

• NMR Relaxation: The underlying theory for relaxation rate predictions.

  Lipari, G., & Szabo, A. (1982). Journal of the American Chemical Society, 104(17), 4546–4559.

• Residual Dipolar Couplings: Seminal work on applying RDCs to proteins.

  Bax, A., & Tjandra, N. (1997). Journal of the American Chemical Society, 119(49), 12041-12042.

• Nuclear Overhauser Effect: Foundational experimental observation.

  Solomon, I. (1955). Physical Review, 99(2), 559.

• 2D NOESY: Development of two-dimensional NOE spectroscopy for biomolecules.

  Kumar, A., Ernst, R. R., & Wüthrich, K. (1980). Biochemical and Biophysical Research Communications, 95(1), 1-6.

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

Support

For issues and questions, please use the GitHub issue tracker.