A swarm intelligence framework for reconstructing gene networks: searching for biologically plausible architectures

Area of Science:

• Computational Biology
• Systems Biology
• Bioinformatics

Background:

• Gene regulatory networks (GRNs) control cellular functions.
• Reverse engineering GRNs from temporal gene expression data is crucial for understanding biological systems.
• Existing methods often struggle with the complexity and scale of GRNs.

Purpose of the Study:

• To develop a novel computational intelligence framework for reverse engineering GRN topology.
• To improve the accuracy and biological plausibility of reconstructed GRNs.
• To leverage swarm intelligence and recurrent neural networks for GRN modeling.

Main Methods:

• Utilized Ant Colony Optimization (ACO) for discrete search of network architectures.
• Employed Particle Swarm Optimization (PSO) for continuous parameter optimization of Recurrent Neural Networks (RNNs).
• Introduced a novel ACO-based solution construction process to generate biologically plausible candidate architectures.

Main Results:

• Successfully reconstructed a small artificial gene network.
• Accurately reverse engineered a noisy synthetic dataset of yeast (S. cerevisiae) gene interactions.
• Effectively reverse engineered the SOS response system in Escherichia coli using real-world data.

Conclusions:

• The proposed framework demonstrates the advantage of incorporating biological knowledge into the search process.
• Problem-specific knowledge significantly enhances the feasibility and accuracy of GRN reconstruction.
• This hybrid approach offers a powerful tool for dissecting complex gene regulatory systems.