Expression profiles of metabolic models to predict compartmentation of enzymes in multi-compartmental systems

Abstract

Enzymes and other proteins coded by nuclear genes are targeted towards various compartments in the plant cell. Here, we describe a method by which localisation of enzymes in a plant cell may be predicted based on their transcription profile in conjunction with analysis of the structure of the metabolic network. This method uses reaction correlation coefficients to identify reactions in a metabolic model that carry similar flux. First a correlation matrix for the expression of genes of interest is calculated and the columns clustered hierarchically using the correlation coefficient. The rows clustered using reaction correlation coefficients. In the resulting matrix, we show that the genes in a particular compartment are clustered together and compartmental predictions, with respect to a reference gene can be readily made.