nmrtoolbox 7.2

No project description provided

nmrtoolbox

Introduction

This is a simple utility that provides modules for working with NMRPipe peak tables and performing a receiver operator characteristic (ROC) analysis to quantify the quality of the "recovered" peaks relative to a control set of "injected" peaks. The modules in this package are as follows:

• nmrtoolbox.peak: classes for reading in peak tables (currently supports both synthetically generated and recovered peak tables from NMRPipe)
• nmrtoolbox.roc: perform receiver operator characteristic (ROC) analysis of a recovered peak table relative to an injected peak table
• nmrtoolbox.mask: define regions of a spectrum that contain signal or are empty
• nmrtoolbox.util: various supporting utilities used by other modules

Applications

Example #1 - Formal Workflow

The tools in this package are utilized by the NUScon software package. You can access NUScon on the NMRbox platform (free for academic, government, and non-profit users). Running nuscon -h will provide instructions on how to run the NUScon evaluation workflow, which directly utilizes the tools presented here in the nmrtoolbox package.

Example #2 - Kick the Tires

from nmrtoolbox.peak import PeakTablePipeRec, PeakTablePipeInj
from nmrtoolbox.roc import roc

# read a peak table generated by NMRPipe peak picker
pt_rec = PeakTablePipeRec(<file-recovered.tab>)
# remove peaks not in the specified NMRPipe classification
pt.reduce(cluster_type=1)
# remove peaks whose linewidths are outliers
pt.reduce(chi2prob=.75)

# read a peak table generated by NMRPipe>genSimTab.tcl 
pt_inj = PeakTablePipeInj(<file-injected.tab>)

# perform ROC analysis
my_roc = roc(
    recovered_table=pt_rec,
    injected_table=pt_inj,
)

# show and plot results
my_roc.print_stats()
my_roc.plot_roc()
my_roc.plot_peaks()
my_roc.plot_outliers()

The nmrtoolbox.peak.reduce() function supports the following filter criteria:

• number
• height
• abs_height
• roi
• index
• cluster_type
• chi2prob
• mask

Note: Filtering by mask requires the external use of NMRPipe to generate a mask file indicating where the spectrum is empty. This binary data is converted by Connjur Spectrum Translator into a "tabular" format file (i.e. plain text) which is then read in by nmrtoolbox.mask.

Changelog

v7

• addition of roc module
• addition of mask module

v6

• rename package as nmrtoolbox
• use subclasses to handle NMRPipe peak tables coming from genSimTab or from the peak picker

v5

• new Axis class for containing metadata from peak table header