Integration of metabolic databases for the reconstruction of genome-scale metabolic networks

Karin Radrich,Yoshimasa Tsuruoka,Albert Gevorgyan,Paul Dobson,Jean-Marc Schwartz,Neil Swainston,Gino Baart

doi:10.1186/1752-0509-4-114

Abstract

BackgroundGenome-scale metabolic reconstructions have been recognised as a valuable tool for a variety of applications ranging from metabolic engineering to evolutionary studies. However, the reconstruction of such networks remains an arduous process requiring a high level of human intervention. This process is further complicated by occurrences of missing or conflicting information and the absence of common annotation standards between different data sources.ResultsIn this article, we report a semi-automated methodology aimed at streamlining the process of metabolic network reconstruction by enabling the integration of different genome-wide databases of metabolic reactions. We present results obtained by applying this methodology to the metabolic network of the plant Arabidopsis thaliana. A systematic comparison of compounds and reactions between two genome-wide databases allowed us to obtain a high-quality core consensus reconstruction, which was validated for stoichiometric consistency. A lower level of consensus led to a larger reconstruction, which has a lower quality standard but provides a baseline for further manual curation.ConclusionThis semi-automated methodology may be applied to other organisms and help to streamline the process of genome-scale network reconstruction in order to accelerate the transfer of such models to applications.

Highlights

Genome-scale metabolic reconstructions have been recognised as a valuable tool for a variety of applications ranging from metabolic engineering to evolutionary studies
Common sources of problems are the non-uniqueness of metabolite identifiers, unbalanced atomic species arising from an incorrect stoichiometry or formula for one or more reactants, incorrect or missing cofactors, and enzymes catalysing more than one reaction [15]
We present an original semi-automated methodology aimed at accelerating the process of metabolic network reconstruction, while at the same time avoiding the loss of accuracy consecutive to a fully automated reconstruction

Summary

Introduction

Genome-scale metabolic reconstructions have been recognised as a valuable tool for a variety of applications ranging from metabolic engineering to evolutionary studies. The reconstruction of such networks remains an arduous process requiring a high level of human intervention. This process is further complicated by occurrences of missing or conflicting information and the absence of common annotation standards between different data sources. The reconstruction of the genome-scale metabolic network of an organism represents a major milestone toward better understanding of its properties. While metabolic pathways are convenient abstractions to represent routes of biochemical conversions of small molecules in an organism, their definition is often arbitrary and varies between sources [2]. The process of reconstructing the genomescale metabolic network of an organism remains very labour-intensive. Additional problems are caused by the lack of usage of standards for the annotation of metabolites and reactions, making the comparison of different models extremely difficult

Objectives

Methods

Results

Discussion

Conclusion