How epfl-lcsb/etfl works
Project type and purpose
• ETFL is a domain-specific Python library (an open-source tool) for building and solving constraint-based metabolic models that integrate expression, thermodynamics and resource allocation.
• Its purpose is to let users load metabolic networks (e.g., E. coli iJO1366), augment them with multi-omics constraints via pyTFA/COBRApy/optlang, and perform analyses (MOMA, dynamic FBA, etc.).
File structure
• Top-level directories:
• etfl/ ← main Python package
– core/ ← core model construction (allocations, macromolecules, thermodynamics)
– analysis/ ← high-level analysis routines (dynamic, MOMA, MoPA, summary)
– optim/ ← optimization layer (config, constraints, variables)
– io/ ← serialization (JSON, dict), model import/export
– integration/ ← multi-omics integration (transcriptomics)
– utils/ ← calibration, parsing, general helpers
– data/ ← built-in organism/model data loaders (e.g., E. coli)
– debugging/ ← debug tools for model inspection
– tests/ ← unit tests for each component
• docker/ ← Dockerfiles & scripts to build/run containerized environment
• doc/ ← Sphinx documentation source
• models/ ← static model files (iJO1366.json)
• organism_data/ ← raw and preprocessed organism-specific datasets
• tutorials/ ← example scripts demonstrating usage
• work/ ← experimental code and data (not core)
• setup.py, setup.cfg, LICENSE, README.rst
README for additional insights
• Lists external dependencies: pyTFA, COBRApy, optlang, CPLEX/Gurobi/GLPK.
• Recommends Docker for environment consistency.
• Emphasizes model building → solver interface → analysis workflow.
Important reminders
• Be exhaustive: include every top-level module and its role
• Flow of data & control from start (model load) to finish (result analysis)
• Extensibility: where a user could plug in a new analysis or solver
• Keep component granularity aligned with the file structure (each submodule = one block)