Comparative genomics to improve bacterial metabolic networks reconstruction and functional modelling

Genome-scale network reconstructions are useful tools for understanding cellular metabolism, enabling a concise mathematical representation of an organism's biochemical capabilities. A number of methods and specific databases currently exist to automatically reconstruct a metabolic model of an organism of interest using its genome as input. Nevertheless, these models are usually incomplete and a strong effort in manual refinement is always required to obtain a model allowing the exploration and the characterization of the possible metabolic flux states of an organism. To reduce the gap between automatic generation of a metabolic model and its possible exploitation in the context of functional modeling (e.g. Flux Balance Analysis) we have developed a computational pipeline aimed at integrating genomics data with evidence resulting from multiple metabolic networks comparisons. Through a comparative genomics framework that integrates several sources of genomics-based metrics (sequence to gene to operon to genome multi scale structure, constraints and organization) the pipeline is able to i) advice on missing information (gaps) which are commonly found after draft reconstructions of metabolic models and ii) to provide an insightful and interactive curation guideline. By bridging automatically reconstructed metabolic models and their corresponding draft genomes assemblies, the pipeline is able to provide valuable evolutionary, phenotypic variance and biotechnology pertinent hints helping to complete the procedures. The overall computational approach will be presented together with its application to two case studies, i.e. oil degrading Acinetobacter venetianus strains and antibiotic producing Pseudolalteromonas representatives.