A neuro-evolution approach to infer a Boolean network from time-series gene expressions

Area of Science:

• Systems Biology
• Computational Biology
• Bioinformatics

Background:

• Inferring regulatory networks from time-series gene expression data is crucial but challenging, especially for large-scale networks.
• Existing methods, including genetic algorithm-based Boolean network inference (GABNI), face computational inefficiencies and limitations in representing regulatory functions.
• There is a need for improved methods to accurately and efficiently infer complex gene regulatory networks.

Purpose of the Study:

• To develop a novel and efficient method for inferring large-scale Boolean regulatory networks from time-series gene expression data.
• To improve the representation of regulatory functions compared to existing approaches like GABNI.
• To enhance the accuracy and computational performance in network inference.

Main Methods:

• A novel genetic algorithm integrated with a neural network was developed for Boolean network inference.
• Neural networks were employed to represent regulatory functions, overcoming limitations of Boolean truth tables.
• The method incorporated an extended range for the time-step lag parameter for flexible regulatory function representation.

Main Results:

• The proposed method significantly outperformed five well-known existing methods, including GABNI, in extensive simulations.
• Superior performance was observed in both structural and dynamics accuracy of inferred networks.
• The new approach demonstrated enhanced capability in inferring large-scale Boolean regulatory networks.

Conclusions:

• The novel neural network-based genetic algorithm is a promising tool for inferring large-scale Boolean regulatory networks.
• The method offers improved accuracy and flexibility in representing gene regulatory functions.
• This approach advances the field of systems biology by enabling more effective analysis of gene expression data.