Network inference from perturbation time course data.

Area of Science:

• Systems Biology
• Network Science
• Computational Biology

Background:

• Biological systems are complex networks across scales.
• Identifying network structures from experimental data is challenging.
• Current methods struggle with dynamic behavior, feedback, and noise.

Purpose of the Study:

• To develop a method for robustly inferring biological network structures.
• To specify the experimental data requirements for network reconstruction.
• To address limitations of existing network inference techniques.

Main Methods:

• Integration of a dynamic least squares framework with modular response analysis (DL-MRA).
• Specification of sufficient experimental perturbation time course data.
• Evaluation of DL-MRA for two and three-node networks.

Main Results:

• DL-MRA robustly infers network properties like edge sign, directionality, and feedback loops.
• The approach accounts for dynamic behavior, external edges, and experimental noise.
• Incomplete gene knockdown can be more informative than full knockout for gene regulatory networks.

Conclusions:

• DL-MRA provides a rational basis for experimental data requirements in network reconstruction.
• The method is applicable to cell state transition, intracellular signaling, and gene regulatory networks.
• Robust inference of signaling networks requires restricted conditions; gene regulatory network strategies may need adjustment.