A Hierarchical Framework for Inverse Problems in Biological Cybernetics: a Microbial Community Case Study
Abstract
This is the software snapshot (release v1.0.1) accompanying the paper: "A Hierarchical Framework for Inverse Problems in Biological Cybernetics: Opportunities and Limitations" by J. R. Banga and A. F. Villaverde. It provides a reproducible MATLAB implementation of a four-level hierarchy of inverse problems, illustrated on a synthetic two-strain cross-feeding microbial consortium:- Level 1 — Parameter estimation: multi-start nonlinear least-squares calibration of six kinetic parameters, together with structural identifiability and observability analysis (via STRIKE-GOLDD).- Level 2 — Model discovery: cross-validated sparse (LASSO) recovery of the growth-law structure from window-integrated per-capita growth, with independent held-out validation.- Level 3 — Inverse optimal control: bound-aware inverse-KKT recovery of a centralized community objective, with data-only identifiability ranges and cold-start forward validation.- Level 4 — Inverse differential game: recovery of player-specific objectives from an open-loop Nash equilibrium, certified by a unilateral-deviation test. See README.md for full documentation. Keywords: inverse problems; systems biology; biological cybernetics; parameter estimation; model discovery; sparse regression; inverse optimal control; inverse differential games; identifiability; observability; microbial communities; cross-feeding; MATLAB This version is derived from : https://github.com/cblmbg/inverseproblemsbiocybernetics/tree/v1.0.1 Claude (Anthropic) was used for code analysis and refactoring. Funding: grants PID2023-146275NB-C21 and PID2023-146275NB-C22 funded by MICIU/AEI, 10.13039/501100011033; co-funded by ERDF/EU, DYNAMO-bio project):License: GPLv3
