Upload model
Drop your model here, or choose a file
.xml · .sbml · .json · .mat · up to about 40 MB
No model to hand?
Each stage produces proposals. A proposal reaches the exported model only when you approve it, and every proposal carries the evidence tier it rests on.
Your model is read, curated and written here in this browser tab. It is never uploaded.
Supervised curation of a genome-scale model
Identifiers, duplicate entities, dead ends, blocked reactions, mass and charge balance, energy-generating cycles, reaction directionality, and growth against published laboratory data. Every problem this page finds becomes a single proposal you approve or reject.
You get back SBML, COBRA JSON and MATLAB .mat, a written report of every finding and your decision on it, and a downloadable dashboard of the figures.
What each stage checks
1IdentifiersEvery metabolite and reaction is matched to its BiGG identifier, or to one stable BiGG-style identifier shared by all of its database identifiers. Each proposal is labelled with the evidence it rests on, from an InChIKey down to a name.
2DiscrepanciesTwo identifiers for the same compound in the same compartment are collapsed into one. Two reactions are proposed for merging only when their stoichiometry matches.
3Structural QCDead-end metabolites, reactions with no gene association, and reactions that cannot carry flux even with every exchange open.
4Mass and chargeElement and charge balance judged at the model's own reference charges. A pH slider titrates metabolite charge states for inspection and raises no findings.
5ThermodynamicsReaction directionality checked against published thermodynamic data, plus detection of cycles that make ATP or another energy carrier out of nothing.
6Lab validationYour strain is looked up in GrowthDB, the medium is pulled from MediaDB, and measured growth is set beside what the model predicts. Rates without experimental backing are labelled as presets.