Look back at a tool bridging machine learning and mechanistic models in systems biology

With the release of the Python tool, sbml2hyb, scientists can convert existing Systems Biology Markup Language (SBML) models into hybrid models that integrate traditional mechanistic methods with machine learning (ML). This standalone tool aims to enhance systems biology analysis by combining parametric equations with ML algorithms, particularly useful when mechanistic understanding is limited.

Developed to streamline hybrid model creation, sbml2hyb offers a straightforward export interface and internal format validation, supporting robust model storage back in SBML databases. Complementing this tool is HMOD, a model format that consolidates mechanistic and ML components under the SBML structure. These components can then be reconverted and trained within widely used software, like MATLAB and COPASI.

This advance in hybrid modelling promises to drive further adoption of ML-enhanced systems biology methods, advancing research into complex biological systems. Learn more about the tool’s potential to transform biological model integration and data storage in databases here.

The figure shows the pipeline for SBML-compatible hybrid modeling.

José Pinto, Rafael S Costa, Leonardo Alexandre, João Ramos, Rui Oliveira, SBML2HYB: a Python interface for SBML compatible hybrid modeling, Bioinformatics, Volume 39, Issue 1, January 2023, btad044, https://doi.org/10.1093/bioinformatics/btad044