LegumeGRN: a gene regulatory network prediction server for functional and comparative studies

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• Accurate gene regulatory network (GRN) construction from high-throughput gene expression data remains a significant challenge.
• Advancements in algorithms and increased transcriptomic data availability now make GRN inference more feasible.

Purpose of the Study:

• To develop a user-friendly, web-based computational service for building, analyzing, and visualizing gene regulatory networks.
• To facilitate the investigation of gene regulatory relationships by biologists.

Main Methods:

• Developed a web server integrating transcriptomic and annotation data for model legume species (Medicago truncatula, Lotus japonicus, Glycine max).
• Implemented multiple GRN prediction algorithms: co-expression, Graphical Gaussian Models (GGMs), Context Likelihood of Relatedness (CLR), TIGRESS, and GENIE3.
• Introduced a novel parallel Bayesian network learning algorithm for inferring causal relationships and handling large datasets.

Main Results:

• The web server supports analysis of preloaded legume data and user-uploaded datasets from any organism.
• Users can select specific experiments, genes, and algorithms for flexible GRN analysis.
• The platform enables integrative and comparative analyses of GRNs across different algorithms, experiments, and species.

Conclusions:

• The developed web server provides a flexible and powerful platform for GRN inference and analysis.
• It aids biologists in exploring complex gene regulatory mechanisms across various species.
• The tool supports both standard and causal GRN inference, enhancing biological discovery.