GeNICE: A Novel Framework for Gene Network Inference by Clustering, Exhaustive Search, and Multivariate Analysis.

Area of Science:

• Systems Biology
• Bioinformatics
• Computational Biology

Background:

• Gene network inference from temporal gene expression data is computationally intensive.
• Existing methods struggle with large datasets containing thousands of genes and limited samples.

Purpose of the Study:

• To develop a scalable framework for gene network inference.
• To reduce the computational complexity of gene network inference methods.

Main Methods:

• Proposed GeNICE, a novel framework using probabilistic gene networks and a clustering procedure.
• Grouped genes with similar expression profiles to reduce search space.
• Performed exhaustive search within clusters and multivariate analysis for network simplification.

Main Results:

• GeNICE achieved substantial computational time reduction compared to non-clustering methods.
• Maintained high gene expression prediction accuracy, even with a small number of clusters.
• Achieved ~97% prediction accuracy on a Plasmodium falciparum dataset, revealing scale-free network properties.

Conclusions:

• GeNICE offers a computationally efficient and accurate solution for gene network inference.
• The clustering approach enhances scalability for large-scale biological data.
• The inferred networks exhibit properties consistent with biological systems.