Data-Driven Identification of Rational Nonlinear Dynamics in Biochemical Networks via an Implicit Singular Value...

Area of Science:

• Systems Biology
• Computational Biology
• Biophysics

Background:

• Rational functions are widely used to model biological network dynamics.
• Estimating these dynamics from data is challenging due to complex nonlinearities.
• Existing methods like sparse identification of nonlinear dynamics (SINDy) struggle with rational functions.

Purpose of the Study:

• To develop an effective method for identifying rational form dynamics in complex biological networks.
• To overcome the computational limitations of existing approaches for rational function identification.
• To provide a generalizable tool for analyzing biological system dynamics.

Main Methods:

• Application of singular value decomposition (SVD) to a library of observational functions.
• Mixing state and derivative terms to capture the implicit form of rational functions.
• Utilizing the null space from SVD for model identification, enabling efficient handling of large-scale data.

Main Results:

• Successfully applied the method to four diverse biological models: Michaelis-Menten kinetics, Bacillus subtilis competence network, penicillin production kinetics, and the yeast glycolytic metabolic network.
• Demonstrated the effectiveness and generalizability of the proposed approach.
• Showcased the method's ability to condense data information into the observable space via SVD.

Conclusions:

• The proposed SVD-based method offers an effective and computationally efficient solution for identifying rational dynamics in biological networks.
• This approach overcomes the limitations of traditional methods, particularly for large-scale biological systems.
• The successful application across multiple models highlights its broad applicability in systems biology research.