Deconvolute overlapping NMR peaks
Peakipy - NMR peak integration/deconvolution using python
Simple deconvolution of NMR peaks for extraction of intensities. Provided an NMRPipe format spectrum (2D or Pseudo 3D) and a peak list (NMRPipe, Sparky or Analysis2), overlapped peaks are automatically/interactively clustered and groups of overlapped peaks are fitted together using Gaussian, Lorentzian or Pseudo-Voigt (Gaussian + Lorentzian) lineshape.
The easiest way to install peakipy is with poetry...
cd peakipy; poetry install
If you don't have poetry you can install it with the following command
curl -sSL https://raw.githubusercontent.com/sdispater/poetry/master/get-poetry.py | python
Otherwise refer to the poetry documentation for more details
You can also install peakipy with
setup.py. You will need python3.6 or greater installed.
cd peakipy; python setup.py install
At this point the package should be installed and the main scripts (
peakipy fit and
should have been added to your path.
- Peak list (NMRPipe, Analysis v2.4, Sparky)
- NMRPipe frequency domain dataset (2D or Pseudo 3D)
There are four main commands:
peakipy readconverts your peak list and selects clusters of peaks.
peakipy editis used to check and adjust fit parameters interactively (i.e clusters and mask radii) if initial clustering is not satisfactory.
peakipy fitfits clusters of peaks.
peakipy checkis used to check individual fits or groups of fits and make plots.
You can use the
--help flags for instructions on how to run the programs (e.g. peakipy read -h)
- Pandas DataFrame containing fitted intensities/linewidths/centers etc.
,fit_prefix,assignment,amp,amp_err,center_x,center_y,sigma_x,sigma_y,fraction,clustid,plane,x_radius,y_radius,x_radius_ppm,y_radius_ppm,lineshape,fwhm_x,fwhm_y,center_x_ppm,center_y_ppm,sigma_x_ppm,sigma_y_ppm,fwhm_x_ppm,fwhm_y_ppm,fwhm_x_hz,fwhm_y_hz 0,_None_,None,291803398.52980924,5502183.185104156,158.44747896487527,9.264911100915297,1.1610674220702277,1.160506074898704,0.0,1,0,4.773,3.734,0.035,0.35,G,2.3221348441404555,2.321012149797408,9.336283145411077,129.6698850201278,0.008514304888101518,0.10878688239041588,0.017028609776203036,0.21757376478083176,13.628064792721176,17.645884354478063 1,_None_,None,197443035.67109975,3671708.463467884,158.44747896487527,9.264911100915297,1.1610674220702277,1.160506074898704,0.0,1,1,4.773,3.734,0.035,0.35,G,2.3221348441404555,2.321012149797408,9.336283145411077,129.6698850201278,0.008514304888101518,0.10878688239041588,0.017028609776203036,0.21757376478083176,13.628064792721176,17.645884354478063 etc...
--plot=<path>option selected the first plane of each fit will be plotted in <path> with the files named according to the cluster ID (clustid) of the fit. Adding
plt.show()on each fit so you can see what it looks like. However, using
peakipy checkshould be preferable since plotting the fits during fitting slows down the process a lot.
To plot fits for all planes or interactively check them you can run
peakipy check fits.csv test.ft2 --dims=0,1,2 --clusters=1,10,20 --show --outname=plot.pdf
Will plot clusters 1,10 and 20 showing each plane in an interactive matplotlib window and save the plots to a multipage pdf called plot.pdf. Calling
--first flag results in only the first plane of each fit being plotted.
peakipy check -h for more options.
You can explore the output data conveniently with
In : import pandas as pd In : import matplotlib.pyplot as plt In : data = pd.read_csv("fits.csv") In : groups = data.groupby("assignment") In : for ind, group in groups: ...: plt.errorbar(group.vclist,group.amp,yerr=group.amp_err,fmt="o",label=group.assignment.iloc) ...: plt.legend() ...: plt.show()
Where Gaussian lineshape is
And Lorentzian is
The fit minimises the residuals of the functions in each dimension
Fraction parameter is fraction of Lorentzian lineshape.
The linewidth for the G lineshape is
The linewidth for PV and L lineshapes is
Download from git repo. To test the program for yourself
cd into the
test directory . I wrote some tests for the code itself which should be run from the top directory like so
Comparison with NMRPipe
A sanity check... Peak intensities were fit using the nlinLS program from NMRPipe and compared with the output from peakipy for the same dataset.
Homage to FuDA
If you would rather use FuDA then try running
peakipy read with the
--fuda flag to create a FuDA parameter file
(params.fuda) and peak list (peaks.fuda).
This should hopefully save you some time on configuration.
Thanks to Jonathan Helmus for writing the wonderful
lmfit team for their awesome work.
matplotlib for beautiful plotting.
My colleagues, Rui Huang, Alex Conicella, Enrico Rennella, Rob Harkness and Tae Hun Kim for their extremely helpful input.
Release history Release notifications | RSS feed
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
|Filename, size||File type||Python version||Upload date||Hashes|
|Filename, size peakipy-0.1.30-py3-none-any.whl (67.4 kB)||File type Wheel||Python version py3||Upload date||Hashes View|
|Filename, size peakipy-0.1.30.tar.gz (62.4 kB)||File type Source||Python version None||Upload date||Hashes View|