A simple work flow for biologically inspired model reduction--application to early JAK-STAT signaling

Area of Science:

• Systems biology
• Computational biology
• Biophysics

Background:

• Biological pathway modeling is essential in systems biology.
• Complex models often have numerous unknown parameters, necessitating parameter estimation from experimental data.
• Model identifiability, ensuring parameter estimation validity, depends on model complexity relative to data quantity and quality.

Purpose of the Study:

• To develop and validate a systematic procedure for simplifying biological pathway models to achieve identifiability.
• To address the challenge of unidentifiable models in systems biology by proposing an alternative to parameter fixing.

Main Methods:

• A systematic model simplification procedure involving parameter estimation, identifiability ranking, and iterative structural modification.
• Utilizing multi-start parameter estimations on a parallel cluster for both estimation and variance calculation.
• Employing parameter variances as a criterion for stopping the iterative simplification process.

Main Results:

• The proposed iterative simplification procedure leads to identifiable models.
• Parameter variances serve as effective and interpretable stopping criteria.
• The method focuses on altering model structure rather than fixing parameters.

Conclusions:

• The developed approach successfully simplifies models to achieve identifiability with small parameter variances.
• Applied to the JAK-STAT pathway, the method yielded an identifiable model with an optimal balance between fit and complexity.