Construction and Optimization of a Large Gene Coexpression Network in Maize Using RNA-Seq Data.

Area of Science:

• Plant genomics
• Bioinformatics
• Systems biology

Background:

• Massively parallel sequencing generates vast amounts of genomewide expression data, accelerating maize research.
• Traditional analysis methods may not be optimal for the scale and type of modern expression data.
• Gene coexpression networks (GCNs) are valuable tools for maize gene function prediction and pathway analysis.

Purpose of the Study:

• To evaluate parameters for constructing optimal gene coexpression networks (GCNs) from plant RNA-Seq data.
• To compare different normalization and network inference methods for maize GCNs.
• To assess the impact of sample size and network aggregation on GCN performance.

Main Methods:

• Evaluation of three RNA-Seq data normalization methods.
• Testing of ten network inference methods (six correlation, four mutual information).
• Analysis of 1266 maize samples and application of a ranked aggregation strategy.

Main Results:

• Normalization methods showed similar performance.
• Correlation-based inference methods outperformed mutual information methods for certain genes.
• Increased sample size positively impacted GCN quality.
• Aggregating multiple single networks improved overall performance.

Conclusions:

• The study provides guidelines for building robust maize gene coexpression networks using RNA-Seq data.
• Optimized GCNs enhance the accuracy of gene function prediction and regulatory pathway discovery in maize.
• Network aggregation is a key strategy for improving GCN reliability and performance.