nmrplot: a Python tool for plotting NMR spectra

Learning how to make a proper Python package

A $^{1}H-^{15}N$ spectrum example

TL;DR

nmrplot is a wrapper around nmrglue for plotting NMR spectra using matplotlib. It is also the first proper Python package I make using my own Cookiecutter template with testing, linting, CI, and documentation building.

Motivation

During the last couple of years I have been making extensive use of nmrglue for loading NMR spectra into Jupyter notebooks and preparing figures for presentations and reports.

nmrglue is a great tool for all NMR & Python aficionados: it can read, process and write NMR spectra in most formats. Used with matplotlib, the resulting contour plots are not only nicer than the standard exports from NMR-specific software, but also scriptable.

Example of a $^{13}C$-$^{13}C$ MAS NMR spectrum of model protein SH3 using nmrplot in CLI mode
Example of a $^{13}C$-$^{13}C$ MAS NMR spectrum of model protein SH3 using nmrplot in CLI mode

After consistently using a hacky code-dump module in different NMR research projects, I thought this would be a good opportunity to start a small Python project.

In parallel, I had also been learning some best practices in software development and distribution. My use of version control, documentation and linting was quite lax, and I had never implemented continuous integration. So after some research I put all these parts in a custom cookiecutter (post to come about that) and started the nmrplot project.

Goal

The goal of this tool is generating high quality, information-rich figures directly from processed NMR data. Additional data layers calculated on the fly can be easily implemented. Examples of such additional information include peak analysis — picking, integration, probabilistic assignment —, spectral analysis — signal-to-noise, referencing — or comparison between spectra.

nmrplot is limited to 1D and 2D spectra (or 2D slices of >= 3D spectra). Support for other formats than Bruker — NMRPipe, Varian — should be straightforward, as this is already possible in nmrglue . I guess that is a good opportunity to learn how to use decorators.

Description and use cases

nmrplot is a module with a Command Line Interface (CLI) for direct interaction. The CLI provides basic plotting capabilities from the terminal, while the Spectrum class permits deeper interaction.

Usage from the command line

The CLI is intended to provide quick visualization while navigating a data set. The default parameters are set so running

nmrplot EXPERIMENT_PATH/EXPNO

should give a reasonably good contour plot. The default matplotlib QT backend enables panning, zooming and saving in various formats, among other basic options.

Plotting parameters can be easily tuned by passing options described in the help menu:

nmrplot --help

Usage: nmrplot [OPTIONS] PATH

  Plot a 1D or 2D Bruker NMR spectrum in a given path

Options:
  -p, --pdata TEXT       Number of processing in the Bruker experiment
  -t, --threshold FLOAT  How many times the noise level is the lowest contour
                         above the baseline
  -c, --cmap TEXT        The colormap to use. Options are: viridis, red, blue,
                         green, purple, orange, grey, light_red, light_blue.
                         Only with sign=both: coolwarm
  -n, --nlevs INTEGER    Number of contour levels to draw
  -f, --factor FLOAT     Increment factor between contour levels
  -s, --sign TEXT        Whether to draw positive, negative or both contours.
                         Options are: positive, negative, both
  --help                 Show this message and exit.

I used click to generate this CLI. It is a very convenient tool, as all the help interface is automatically generated from the __main__.py. This, along with the online documentation — also automatically parsed by mkdocstrings — provides the user with complete and handy information with little effort.

Usage in Jupyter notebooks and Python scripts

I find Jupyter notebooks are great environments for making figures and plots for papers and reports, from the initial exploratory stage to the publication stage. Plain scripts can provide the same functionality and are less conflictive with version control, so they are probably a better solution for long-term deposition.

Here is a notebook with an example plot using the Spectrum.plot_spectrum method, but one can also directly access Spectrum.data.

Advantages of nmrplot

Traceability and reproducibility

A notebook or script containing the paths to the data inputs and the figure outputs is a traceable and self-contained solution, perfect for scientific communications. Raw data, processed outputs and analysis can be easily shared with collaborators or made public.

As a task gets more complex than simple on-the-fly analysis, refactoring the code in a separate module and adding environment requirements and dependencies also enforces reproducibility.

Composability

matplotlib is the de facto standard Python library for plotting. It will get the job done for users of any level in most cases, so with a few visits to Stackoverflow, one can compose complex publication figures with multiple subplots, insets, etc.

There also tools in the Python environment that enable interactive plots, like bokeh or plotly. Such format can be interesting for academic researchers in the context of collaboration of dissemination.

Beautiful plots

Most specialized NMR data analysis software I know lack love on the aesthetical side. Nice modern color maps, fonts, alpha compositing, contour level thickness ar little things that can make the difference between a dull plot and a pleasing figure for your presentations and papers.

Example of a $^{1}H$-$^{15}N$ solution NMR spectrum of model protein SH3 using nmrplot in CLI mode
Example of a $^{1}H$-$^{15}N$ solution NMR spectrum of model protein SH3 using nmrplot in CLI mode

Also, exporting options are totally customizable: format, size, resolution… Write the settings once, update all figures whenever you want!

How to install nmrplot

nmrplot can be installed using pip from the command line directly from GitHub:

python -m pip install git+https://github.com/miguelarbesu/nmrplot  

Insiders can install an editable dev copy by cloning or forking the repository and running

python -m pip install -e .
Miguel Arbesú Andrés
Miguel Arbesú Andrés
Researcher in Bio ∩ AI

An organic chemist turned structural biologist, then data scientist.