COBRAGUI

COBRAGUI (COBRA graphical user interface) is the workspace of GEMs MAP that works on the model itself: a graphical front end to COBRApy, the Python package for constraint-based reconstruction and analysis. It lets you load a genome-scale model, look inside it, change its constraints, simulate it and compare results by pointing and clicking. The computations are the standard COBRApy ones, run by a local Python service.

COBRAGUI after importing E. coli iML1515 (BiGG): 2,712 reactions, 1,877 metabolites, 1,516 genes, 3 compartments. Validation notes (dead ends, reactions without gene association, blocked or unbalanced reactions) are listed on the overview and link to the entities involved.

1. Workspace switcher: COBRAGUI, Map Builder, Templates.
2. Sections of COBRAGUI, in the order of a typical analysis: Overview, Explore, Medium, Simulation, Results, Compare, Metabolite analysis.
3. Model state: the scenario whose bounds and objective are used (here the unmodified Base model). Unmodified / Modified · n shows its changes; Structure opens the history of structural edits.
4. Run the selected method on the active state; the arrow chooses another method.
5. Model card: file, format, checksum, size, objective, solver.
6. Inspector: details of whatever you click (metabolite, reaction, gene, subsystem), with back/forward navigation.
7. Status bar: backend and solver, the active flux and FVA results, the numerical zero threshold and the activity log.

Why a graphical interface for COBRApy?

COBRApy is powerful but code-first: inspecting a reaction, changing an uptake rate or comparing two solutions each means writing and re-running code, and the context of a result (which medium, which bounds?) is easily lost. COBRAGUI keeps that context attached to every result and makes the model itself browsable. You can click a metabolite in an equation, see which reactions produce and consume it, and continue to the next one. Nothing replaces COBRApy: the interface sends requests, COBRApy computes, and the answers come back with their provenance.

A typical session

flowchart LR
    I["Import<br/>SBML · JSON · MAT"] --> O["Overview<br/>validation notes"]
    O --> E["Explore<br/>tables · inspector"]
    E --> C["Constrain<br/>medium · bounds · objective"]
    C --> S["Simulate<br/>FBA · pFBA · loopless · FVA"]
    S --> R["Results<br/>tables · provenance"]
    R --> K["Compare<br/>state A vs state B"]
    R --> M["Map Builder<br/>Visualize result"]
    K --> M

1. Import a model and read the validation notes on the overview.
2. Explore metabolites, reactions, genes and subsystems; every entity opens in the inspector.
3. Set the constraints: medium, reaction bounds and objective, kept in named model states. If needed, edit the model structure.
4. Simulate with FBA, pFBA, loopless FBA or FVA.
5. Follow individual metabolites through the solution, and curate their identifiers with MetaNetX.
6. Keep, compare and export results, or send them to the Map Builder.

What is kept, and where

What	Where	Lifetime
The imported model file	Backend cache (a copy)	Until you delete the cache. The original file is never modified.
Model states, results, favorites, saved views, medium presets	Browser (IndexedDB), per model checksum	Restored when the same model file is imported again.
Structural edits and metabolite curation	Backend cache, as logs next to the model	Re-applied whenever that model is loaded.
Exported models and results	Files you save	Permanent; they record the state, edits and curation they include.

Because sessions are keyed by the model's checksum, importing a different version of a model starts a new session.

Keyboard

Keys	Action
Ctrl+K or /	Search the model or run a command
F	Focus the table search (Explore)
R / M / G	Reactions / Metabolites / Genes tab
Alt+Left / Alt+Right	Inspector back / forward
Ctrl+Z / Ctrl+Shift+Z	Undo / redo a model change
Ctrl+Enter	Run the selected method
I	Show or hide the inspector
Alt+1 … Alt+6	Overview, Explore, Medium, Simulation, Results, Compare

Start here: Importing models.