Intervention in gene regulatory networks with maximal phenotype alteration

Area of Science:

• Computational Biology
• Genomics
• Systems Biology

Background:

• Modeling gene interactions via gene regulatory networks (GRNs) is crucial for understanding intervention effects.
• Current methods for shifting GRN steady-state mass lack optimality for long-term behavior.
• Phenotype is viewed as the long-run behavior (steady-state distribution) of the GRN.

Purpose of the Study:

• To develop an optimal control policy for shifting GRN steady-state mass.
• To provide an informatics environment for studying intervention effects and deriving strategies.
• To directly optimize the shift of probability mass from undesirable to desirable states.

Main Methods:

• A linear programming approach is employed for optimal control.
• The method handles both unconstrained and constrained optimization scenarios.
• The theory is applicable to any Markovian GRN, demonstrated with probabilistic Boolean networks.

Main Results:

• The linear programming method optimally shifts steady-state mass, outperforming prior approaches.
• Constrained optimization allows limiting mass shift to ambiguous states with perceived risk.
• The optimization is directly based on the amount of mass shift achieved.

Conclusions:

• Linear programming offers an optimal control policy for GRN interventions.
• This method provides a robust framework for analyzing and manipulating complex biological networks.
• The approach is versatile and applicable to various Markovian GRNs.