Dynamic probabilistic threshold networks to infer signaling pathways from time-course perturbation data

Area of Science:

• Systems Biology
• Computational Biology
• Network Science

Background:

• Molecular network inference is crucial for understanding biological systems.
• Reconstructing these networks from experimental data is challenging due to inherent limitations.
• Prior biological knowledge or regularization is often required for accurate inference.

Purpose of the Study:

• To develop a novel method for inferring signaling networks from time-course perturbation data.
• To utilize dynamic Bayesian networks with probabilistic Boolean threshold functions for modeling protein activation.
• To analyze model posterior distributions for reconstructing molecular interaction networks.

Main Methods:

• Employed dynamic Bayesian networks with probabilistic Boolean threshold functions.
• Utilized evolutionary Markov Chain Monte Carlo (MCMC) sampling for posterior distribution analysis.
• Applied clustering techniques to identify probability distributions over potential networks.

Main Results:

• Evaluated the method on simulated data, assessing performance against dataset size and noise levels.
• Demonstrated superior performance compared to current state-of-the-art methods on simulated datasets.
• Successfully applied the method to infer signaling networks in the EGF-mediated ERBB pathway.

Conclusions:

• Dynamic Probabilistic Threshold Networks offers a robust approach for signaling network reconstruction.
• The method effectively leverages dynamic system responses post-perturbation.
• The approach identified known and predicted novel interactions within the ERBB pathway.