A Relationship Prediction Method for Magnaporthe oryzae-Rice Multi-Omics Data Based on WGCNA and Graph Autoencoder

Area of Science:

• Plant Pathology
• Bioinformatics
• Genomics

Background:

• Magnaporthe oryzae Oryzae (MoO) is a major rice pathogen, but its infection mechanisms are not fully understood.
• Single-omics data is insufficient to elucidate complex cross-kingdom interactions between MoO and rice.
• Existing methods lack the capacity for integrated multi-omics analysis.

Purpose of the Study:

• To develop a novel computational method for integrating multi-omics data from MoO and rice.
• To construct a comprehensive interaction network revealing key biomolecules involved in rice-fungus pathogenesis.
• To provide a foundation for understanding MoO infection and developing disease-resistant rice.

Main Methods:

• Proposed Weighted Gene Autoencoder Multi-Omics Relationship Prediction (WGAEMRP).
• Combined weighted gene co-expression network analysis (WGCNA) with graph autoencoder.
• Applied WGAEMRP to MoO-rice multi-omics data to build a heterogeneous interaction network.

Main Results:

• Constructed a MoO-rice multi-omics heterogeneous interaction network.
• Identified key biomolecules: 18 MoO small RNAs (sRNAs), 17 rice genes, 26 rice mRNAs, and 28 rice proteins.
• Discovered functional modules and pathways related to gene expression, protein dynamics, and metabolism during infection.

Conclusions:

• WGAEMRP significantly improves multi-omics data integration efficiency and accuracy.
• The study provides a robust data foundation for investigating MoO pathogenesis.
• Findings offer new insights for developing novel strategies for disease-resistant rice varieties.