sbml2hyb

sbml2hyb converts SBML-encoded mechanistic models into hybrid semiparametric models that integrate mechanistic functions with machine learning (ML) to support systems biology analysis.

Key Features:

• Conversion Capability: Converts SBML-encoded mechanistic models into hybrid models that combine mechanistic and ML components.
• Implementation: Implemented in Python.
• Validation: Includes an internal format validator for model format consistency.
• HMOD Format Support: Defines and uses the HMOD (Hybrid Model Definition) format to aggregate mechanistic and ML information while adhering to SBML standards.
• Training and Export: Supports training of hybrid models and exporting trained models back into SBML format.
• Case Studies: Demonstrated through two case studies illustrating application to real-world biological models.

Scientific Applications:

• Hybrid systems biology modeling: Enables construction of semiparametric hybrid models that combine mechanistic insight with data-driven ML components.
• Analysis of complex biological phenomena: Facilitates study of systems where purely mechanistic or purely statistical models are insufficient.
• Improved model-based inference: Supports more comprehensive and accurate analyses to aid discovery in systems biology.

Methodology:

Converts SBML-encoded mechanistic models into hybrid semiparametric models by incorporating ML components; aggregates mechanistic and ML information into HMOD consistent with SBML; validates formats with an internal validator; trains hybrid models and exports them back to SBML.