Inferring Drosophila gap gene regulatory network: a parameter sensitivity and perturbation analysis

Area of Science:

• Systems Biology
• Computational Biology
• Genetics

Background:

• Accurate data and system descriptions are crucial for inverse modeling of gene regulatory networks (GRNs) simulating spatio-temporal processes.
• Connectionist models simulate gene expression dynamics but parameter estimation can yield multiple valid circuits, necessitating robustness evaluation.

Purpose of the Study:

• To investigate the sensitivity and robustness of gene regulatory network circuits derived from reverse engineering.
• To address circuit discrimination challenges in inverse modeling of gene regulatory networks.

Main Methods:

• Utilized reverse engineering of a connectionist model to infer gene regulatory networks from spatio-temporal gene expression data.
• Employed least squares minimization for circuit discrimination based on simulation error.
• Introduced stochastic modeling derived from deterministic solutions to analyze circuit robustness.

Main Results:

• Parameter sensitivity analysis successfully discriminated between circuits with similar quantitative behavior but different qualitative properties.
• Stochastic modeling revealed that circuit robustness to fluctuations is modular rather than global.
• A strong correlation was found between circuit sensitivity and robustness to fluctuations.

Conclusions:

• Reverse engineering of GRNs should extend beyond parameter estimation to include model properties like robustness and sensitivity.
• Multi-objective optimization incorporating robustness and sensitivity analysis is recommended for improved GRN inference.
• Circuit robustness in gene regulatory networks is modular, not a global property.