Inferring gene regulatory networks using a time-delayed mass action model.

Area of Science:

• Systems Biology
• Computational Biology
• Biophysics

Background:

• Gene regulatory networks (GRNs) are crucial for cellular functions.
• Modeling GRNs is essential for understanding complex biological processes.
• Existing models often lack the ability to incorporate time delays in gene interactions.

Purpose of the Study:

• To develop a novel time-delayed mass action model for GRNs using delay differential equations (DDEs).
• To propose an efficient learning method for inferring GRN structure and parameters from time-series gene expression data.
• To evaluate the model's performance, including its anti-noise ability, on known GRN motifs and a biological network.

Main Methods:

• Implementation of a time-delayed mass action model using delay differential equations (DDEs).
• Development of a hybrid learning algorithm combining Population-Based Incremental Learning (PBIL) and Trigonometric Differential Evolution (TDE).
• Application of the model and algorithm to synthetic GRN motifs and the budding yeast cell cycle network.

Main Results:

• Successful inference of GRN structures and parameters from time-series gene expression data.
• Demonstration of strong anti-noise capabilities in the proposed model.
• Significant performance improvements compared to existing methods in GRN modeling.

Conclusions:

• The time-delayed mass action model effectively captures GRN dynamics, including gene interaction delays.
• The proposed PBIL-TDE learning method efficiently infers GRN topology and parameters.
• This approach offers a robust and accurate framework for GRN analysis and modeling.