A multi-approaches-guided genetic algorithm with application to operon prediction.

Area of Science:

• Genomics
• Bioinformatics
• Systems Biology

Background:

• Operon prediction is crucial for understanding bacterial gene regulation and reconstructing genome-wide regulatory networks.
• Existing methods often use single approaches for multiple genome features, limiting the exploitation of unique data characteristics.

Purpose of the Study:

• To develop a combined method for operon prediction that utilizes different preprocessing techniques for distinct genome features.
• To leverage the unique characteristics of intergenic distance, gene function, metabolic pathways, and gene expression data for improved operon prediction.

Main Methods:

• A multi-approach-guided genetic algorithm was developed for operon prediction.
• Intergenic distance was processed using a novel local-entropy-minimization method.
• Log-likelihood for Cluster of Orthologous Groups (COG) gene functions and Pearson correlation coefficient for microarray expression data were calculated.
• A genetic algorithm integrated these four data types: intergenic distance, COG functions, metabolic pathways, and microarray expression.

Main Results:

• The method was evaluated on Escherichia coli, Bacillus subtilis, and Pseudomonas aeruginosa genomes.
• Prediction accuracies achieved were 85.9987% for E. coli, 88.296% for B. subtilis, and 81.2384% for P. aeruginosa.
• The approach demonstrated high predictive performance across diverse prokaryotic genomes.

Conclusions:

• Preprocessing genome data with multiple approaches within a genetic algorithm effectively utilizes diverse biological characteristics.
• The developed method is applicable and effective for operon prediction in prokaryotes.
• This approach facilitates more accurate reconstruction of gene regulatory networks.