Granger causality in integrated GC–MS and LC–MS metabolomics data reveals the interface of primary and secondary metabolism

• DOI: 10.1007/s11306-012-0470-0
• Publication Date: 2012-10-24T09:04:52Z
• Authors: Hannes Doerfler; David Lyon; Thomas Nägele; Xiaoliang Sun; Lena Fragner; Franz Hadacek; Volker Egelhofer; Wolfram Weckwerth
• Keywords: 106031 Plant physiology; 1040 Chemie; 0301 basic medicine; Endocrinology, Diabetes and Metabolism; 106002 Biochemie; Clinical Biochemistry; 106031 Pflanzenphysiologie; 106002 Biochemistry; Biochemistry; 03 medical and health sciences; 106007 Biostatistik; Original Article; 1040 Chemistry; 106007 Biostatistics

Metabolomics has emerged as a key technique of modern life sciences in recent years. Two major techniques for metabolomics in the last 10 years are gas chromatography coupled to mass spectrometry (GC-MS) and liquid chromatography coupled to mass spectrometry (LC-MS). Each platform has a specific performance detecting subsets of metabolites. GC-MS in combination with derivatisation has a preference for small polar metabolites covering primary metabolism. In contrast, reversed phase LC-MS covers large hydrophobic metabolites predominant in secondary metabolism. Here, we present an integrative metabolomics platform providing a mean to reveal the interaction of primary and secondary metabolism in plants and other organisms. The strategy combines GC-MS and LC-MS analysis of the same sample, a novel alignment tool MetMAX and a statistical toolbox COVAIN for data integration and linkage of Granger Causality with metabolic modelling. For metabolic modelling we have implemented the combined GC-LC-MS metabolomics data covariance matrix and a stoichiometric matrix of the underlying biochemical reaction network. The changes in biochemical regulation are expressed as differential Jacobian matrices. Applying the Granger causality, a subset of secondary metabolites was detected with significant correlations to primary metabolites such as sugars and amino acids. These metabolic subsets were compiled into a stoichiometric matrix N. Using N the inverse calculation of a differential Jacobian J from metabolomics data was possible. Key points of regulation at the interface of primary and secondary metabolism were identified.