Genomic data assimilation for estimating hybrid functional Petri net from time-course gene expression data

Area of Science:

• Computational Biology
• Systems Biology
• Bioinformatics

Background:

• Traditional methods for constructing biological pathway simulation models rely on empirical tuning of parameters and network structure.
• This empirical approach has limitations in scalability for complex biological networks.

Purpose of the Study:

• To propose an automatic construction method for hybrid functional Petri net models of biological pathways.
• To extend the capability of simulation models by employing a data assimilation approach.

Main Methods:

• Utilized a genomic data assimilation framework linking simulation models with observed data (e.g., microarray gene expression).
• Employed a nonlinear state space model where unknown simulation parameters are treated as parameters of the state space model.
• Applied maximum a posteriori (MAP) estimators for parameter estimation within a Bayesian inference framework.

Main Results:

• Demonstrated the effectiveness of the proposed approach using synthetic data.
• Achieved successful parameter estimation and suitable network structure selection through genomic data assimilation.
• Showcased the ability to handle both model construction and parameter tuning within a unified Bayesian framework.

Conclusions:

• The proposed genomic data assimilation framework offers an effective solution for automatic construction and parameter tuning of biological pathway simulation models.
• Bayesian inference within this framework helps control overfitting, leading to more reliable biological pathway models.
• This approach overcomes the limitations of empirical methods, enabling the simulation of larger and more complex biological networks.