GEM Inspector is a curation and debugging toolkit for genome-scale metabolic models (GEMs).
Curating a genome-scale metabolic model is rarely a straightforward process. Models can contain blocked reactions, dead-end metabolites, unexpected essential nutrients, missing pathways, inconsistent annotations, energy-generating cycles, and other issues that are often difficult to identify and trace back to their source. Debugging these problems can also become repetitive, especially when working with multiple models.
GEM Inspector is being developed to make this process easier. It brings together model inspection, diagnostics, comparative analyses, and debugging tools in one place, with a particular focus on why a model behaves the way it does and where curation may be needed.
The tool is currently a work in progress, and we are actively expanding and improving its functionality based on ongoing use and feedback.
Current status: GEM Inspector currently runs locally. You will need to clone this repository and run the application on your machine - supports windows, linux and mac (Installation & Setup). We are also exploring the possibility of hosting it as an online tool in the future.
The current version contains four main modules, covering different aspects of GEM inspection and curation:
Model's overall structure and key properties.
- Model information and metadata
- Genes, metabolites, reactions, and compartment statistics
- Objective function and growth information
- Reaction and metabolite browsing
- Exchange reactions overview
- Diet/media selection
- MEMOTE based quality assessment
- Minimal medium composition analysis
Tools for identifying common structural and functional issues and comparing model behaviour.
- Essential exchange analysis
- Essential gene and reaction analysis
- Blocked reaction detection
- Dead-end metabolite detection
- Demand and sink audits
- Network gap analysis
- FROG reproducibility analysis
- Model-level reports and CSV exports
The debugging module is a major focus of the tool. It is designed to help move beyond simply identifying an issue and instead trace and investigate what is happening inside the model.
- Essential exchange debugging
- Biomass precursor production checks
- Pathway tracing
- Before/after knockout comparisons
- Flux-pathway visualisation
- Energy-cycle and futile-loop investigation
- Currency/cofactor-aware pathway tracing
- Interactive network views
- Debugging and comparison reports
- Compare two models: naming conventions, then genes, metabolites, reactions and biomass components shared or unique to each, with copy / CSV export of ids and MetaNetX ids
Tools for working with model annotations and translating information across identifier systems.
- Annotation coverage
- Identifier and namespace inspection
- Annotation translation across namespaces
- Model annotation quality checks
- SEED to VMH translation using the tables in
data/(mapped / errors / no available mapping, each exportable), with the translated model and report downloadable as a zip
GEM Inspector currently accepts:
- SBML / XML
- JSON
- YAML
- MATLAB
- Supported compressed SBML files
Essential exchanges can be investigated individually by tracing active-flux pathways from the exchanged metabolite towards biomass.
The tool allows comparison of:
- Before KO — the pathway in the original FBA solution
- After KO — the pathway after knocking out the exchange
This helps identify whether the exchange is required because it feeds a specific pathway, whether an alternative route exists, and where the effect of the knockout propagates through the model.
The general-purpose pathway tracer allows users to trace active metabolic routes between metabolites.
It supports:
- Metabolite- or reaction-based starting points
- Configurable tracing depth
- Multiple possible paths
- Currency/cofactor filtering
- Text-based pathway views
- Interactive Cytoscape.js network visualisation
- Before/after knockout comparisons
The network audit tools provide complementary views of model gaps:
- Blocked reactions — reactions unable to carry flux
- Dead-end metabolites — metabolites that can only be produced or only consumed
- Demand & sink reactions — usage and essentiality of demand/sink reactions
These analyses can be independently exported as CSV files.
GEM Inspector allows users to select or define the medium used for analysis, including:
- Original model bounds
- Complete media
- Minimal media
- Custom diets
A separate minimal-medium analysis can also determine the smallest set of nutrients required to achieve a selected growth level.
The FROG module provides model reproducibility checks through:
- F — Flux Variability Analysis
- R — Reaction deletions
- O — Objective value
- G — Gene deletions
Results can be inspected within the application and exported as CSV or a standalone HTML report.
Results from multiple modules can be combined into a single self-contained HTML report for documentation, comparison, and sharing.
GEM Inspector is currently designed to be run locally.
Clone this repository and follow the installation instructions:
chmod +x run.sh
./run.shrun.batOnce the application starts, open:
http://127.0.0.1:5000
For the required Python environment and dependencies, see requirements.txt.
GEM Inspector is built using tools and libraries from the GEM/constraint-based modelling ecosystem, including COBRApy, MEMOTE, and Cytoscape.js, and was developed with the assistance of Claude (Anthropic).
For more information on the implementation and individual analyses, see the documentation and source code in this repository.