Challenges in experimental data integration within genome-scale metabolic models

The meeting "Challenges in experimental data integration within genome-scale metabolic models" was held at the Institut Henri Poincaré, Université Pierre et Marie Curie, Paris, October 10th and 11th, 2009 [1]. It brought together leading international researchers in the field of genome-scale metabolic modelling and enzyme-kinetics modelling. As suggested by the title, the emphasis was on innovative methodologies aimed at taking better advantage of various experimental data types (such as measurements of flux and intra-cellular metabolite concentrations, tracing of isotopomers, mutant growth phenotypes and gene expression datasets). These kinds of data will increasingly empower researchers aiming to characterize metabolism in various biological systems, as well as its evolution. In this report, we outline the most important advances presented at the meeting.While the number of fully sequenced genomes continues to grow at an exponential rate, the number of published reconstructions of metabolic models [2] is dramatically lagging behind the sequencing effort. This slow pace of model reconstruction effort was highlighted by both David Fell (Oxford Brookes University, UK) and Costas Maranas (Penn State University, USA) at the meeting. While various automatic procedures have been introduced during this past decade to assist the reconstruction of metabolic models, their output still requires a painstaking curation effort. Fell discussed various kinds of inconsistencies that are prevalent in many existing genome-scale metabolic reconstructions including presence of dead-end metabolites, stoichiometric imbalance of certain reactions and erroneous reaction directionality assignments [3]. He also stressed the need to develop automated heuristics for both fast supervised curation of existing models and for the construction of new metabolic models. Instances of such methods were presented by Maranas, who developed with his colleagues novel algorithms including GapFill and GapFin