Collocation-based sparse estimation for constructing dynamic gene networks.

Area of Science:

• Systems Biology
• Computational Biology
• Bioinformatics

Background:

• Inferring dynamic gene networks is crucial for understanding biological processes.
• Existing models like Bayesian networks assume stationary time lags and struggle to separate process and measurement noise.
• These limitations hinder accurate analysis of time-series gene expression data.

Purpose of the Study:

• To propose a novel stochastic differential equation model for inferring continuous-time dynamic gene networks.
• To address the challenges of stationary time lags and inseparable noise components in gene network inference.
• To develop a computationally efficient method for parameter estimation and model selection.

Main Methods:

• A stochastic differential equation model is developed to capture continuous-time gene dynamics with both process and observation noise.
• A collocation-based sparse estimation technique is employed for simultaneous parameter estimation and model selection.
• Biological knowledge is integrated to enhance estimation accuracy.

Main Results:

• The proposed method effectively infers continuous-time dynamic gene networks.
• The collocation-based approach demonstrates reduced computational requirements compared to traditional methods.
• Analysis of human primary small airway epithelial cells revealed significant genes influenced by gefitinib, outperforming existing approaches.

Conclusions:

• The developed stochastic differential equation model offers a robust framework for dynamic gene network inference.
• The collocation-based sparse estimation provides an efficient and accurate method for analyzing complex biological systems.
• This approach enhances our ability to identify drug-affected genes and understand cellular responses.