Sequel improves on original: Analyzing plant metabolomics data with PARAFAC2

Skip to Navigation


  • Published: Nov 5, 2012
  • Author: Jon Evans
  • Channels: Laboratory Informatics
thumbnail image: Sequel improves on original: Analyzing plant metabolomics data with PARAFAC2

Three-way datasets

Sequel improves on original: Analyzing plant metabolomics data with PARAFAC2

As a demonstration of the power of the data processing technique known as PARAFAC2 (Parallel Factor analysis 2), Danish scientists have used it to more than double the number of metabolites associated with the natural insect resistance of the herb rocketcress (Barbarea vulgaris).

Like the original PARAFAC, PARAFAC2 is an extension of principal component analysis (PCA) from two-way datasets, comprising a single two dimensional matrix, to three-way datasets, comprising multiple two dimensional matrices. Such three-way datasets are commonly produced by the analysis of multiple samples by chromatography and mass spectrometry, which generate a matrix of mass-to-charge ratios against retention times for each sample.

PARAFAC and PARAFAC2 both decompose this kind of three-way dataset into a more compact form, in the process resolving all the chromatographic peaks in the data. They do this by converting the original data into a scores matrix and two loading matrices. The scores matrix represents the concentration profiles of the resolved peaks, while the two loading matrices represent the elution time profiles and the mass spectral profiles.


Aligning peaks

PARAFAC works by combining the elution profiles and mass spectral profiles for all the samples, generating a single, common elution profile and mass spectral profile. In contrast, it determines a separate score matrix for each sample. In this way, it can detect common analytes across all the samples, while determining the concentrations of each analyte in each sample. But to work properly the peaks have to be perfectly aligned across all the samples, which usually means that some form of alignment technique, such as correlation optimized warping, needs to be applied to the data before it can be processed by PARAFAC.

PARAFAC2 does away with this alignment step, as it can work with peaks that aren’t perfectly aligned between samples. To do this, rather than generating a common elution profile, it generates separate elution profiles for each of the samples and then aligns the peaks in these profiles across the different samples by matching up the mass spectral profiles. So far, PARAFAC2 has mainly been used to process data produced by gas chromatography-liquid chromatography, but now scientists led by Søren Balling Engelsen at the University of Copenhagen have tried it out on plant metabolomics data generated by LC-MS.


Classifying metabolites

This data was generated by an earlier study looking for the metabolites responsible for the natural resistance to an agricultural pest known as the flea beetle (Phyllotreta nemorum) found in rocketcress, which tends to be toxic to flea beetle larvae that feed on it. In the original study, 127 rocketcress plants that differed in their level of resistance were analysed by LC-MS. The resultant data were then processed using PCA, which uncovered four triterpenoid saponins that appeared to be associated with increased levels of resistance to the flea beetle larvae.

When Engelsen and his colleagues analysed this data using PARAFAC2, not only did they find the same four triterpenoid saponins, but they also found five other saponin-like metabolites that appeared to be associated with levels of resistance. Three of these new metabolites were associated with high levels of resistance, but two of them were associated with low levels of resistance.

Engleson and his colleagues were unable to determine fully the identity or chemical composition of these five new metabolites, but the fact they were found at all shows that PARAFAC2 offers a powerful new way to process complex LC-MS data.

Related Links

Journal of Chromatography A (Article in Press): "Plant metabolomics: Resolution and quantification of elusive peaks in liquid chromatography–mass spectrometry profiles of complex plant extracts using multi-way decomposition methods"

Article by Jon Evans

The views represented in this article are solely those of the author and do not necessarily represent those of John Wiley and Sons, Ltd.

Social Links

Share This Links

Bookmark and Share


Suppliers Selection
Societies Selection

Banner Ad

Click here to see
all job opportunities

Most Viewed

Copyright Information

Interested in spectroscopy? Visit our sister site

Copyright © 2019 John Wiley & Sons, Inc. All Rights Reserved