Network inference via adaptive optimal design.

Area of Science:

• Systems Biology
• Computational Biology
• Network Science

Background:

• Current network reverse engineering methods for genetic and metabolic networks often lack rigorous experimental design.
• This deficiency leads to uncertainties in data and knowledge, impacting parameter estimates and network reconstruction reliability.

Purpose of the Study:

• To propose and evaluate a novel method for iterative experimental design in network reverse engineering.
• To address the analysis and propagation of uncertainties in data and prior knowledge.
• To improve the reliability of inferred network topology and parameter estimates.

Main Methods:

• An iterative experimental design approach is presented, adapting based on the most recent available data.
• The method explicitly incorporates the analysis and propagation of uncertainties from both data and existing knowledge.
• Demonstrated on genetic (mRNA synthesis/degradation) and molecular (cAMP signaling) networks.

Main Results:

• Substantial reduction in parameter uncertainty was achieved for both small-scale test networks.
• The proposed method demonstrates improved reliability in network reconstruction.
• Extension to larger networks requires more rigorous parameter estimation algorithms incorporating sparsity.

Conclusions:

• Careful experimental design is crucial for enhancing the reliability of inferred network topology.
• For large-scale networks, advanced parameter estimation algorithms with sparsity constraints are necessary.
• The adaptive optimal design setting, as demonstrated, is applicable to complex biological networks.