The RAVEN Toolbox and Its Use for Generating a Genome-scale Metabolic Model for Penicillium chrysogenum
Figure 3
Overview of the genes which are unique to the automatically reconstructed model and iIN800, respectively.
Saccharomyces Genome Database was used to classify the genes. Green corresponds to genes where the function is well-defined and suited for GEMs, basically enzymes involved in metabolism. Red corresponds to genes where the function is unknown, where the corresponding protein is not an enzyme or where the function is in signaling rather than metabolism. These genes should normally not be present in a GEM. Blue corresponds to genes that are putative enzymes or where the ORF is a functional enzyme in some strains but not in others. As can be seen, the automatically reconstructed model has both a larger number of unique genes and a larger proportion of enzymes compared to the published model. For iIN800 some enzymatic genes are further classified as “polymer”, “lipid” or “membrane”. These are parts of metabolism where an automatically generated model from KEGG would have particular drawbacks compared to a manually reconstructed model. “Polymer” corresponds mainly to genes involved in sugar polymer metabolism, which is an area that contains many unbalanced reactions in KEGG. Such reactions were excluded when the validation model was generated, so the corresponding genes could not be included. The same holds for “lipid”, where the reactions contain many general metabolites. This also results in excluded reactions. “Membrane” corresponds to reactions which depend on one metabolite but in two different compartments. This compartmentalization information is absent in KEGG so the equation becomes incorrect and it is therefore excluded.