In silico annotation of discriminative markers of three Zanthoxylum species using molecular network derived annotation propagation

Abstract

In liquid chromatography-mass spectrometry (LC-MS) metabolomics, data matrices with up to thousands of variables for each ion peak are subjected to multivariate analysis (MVA) to assess the homogeneity between samples. The large dimensions of LC/MS datasets hinder the identification of the discriminant or the metabolic markers. In the present study, the molecular network (MN) approach and two in silico annotation tools, network annotation propagation (NAP) and the hierarchical chemical classification method, ClassyFire, were used to annotate the metabolites of three Zanthoxylum species, Z. bungeanum, Z. schinifolium and Z. piperitum. The in silico annotation results of the MN nodes and the MVA variables were combined and visualized in loading plots. This approach helped intuitive detection of the variables that greatly contributed to the separation of the samples in the score plot as discriminant or metabolic markers, thereby allowing rapid annotation of two flavanone derivatives.