MiDNE a tool for Multi-omics genes and drugs interactions discovery

Area of Science:

• Computational Biology
• Bioinformatics
• Systems Biology

Background:

• Understanding complex diseases requires models of molecular interactions.
• High-throughput omics technologies enable disease-specific modeling.
• Analyzing single omics layers misses crucial cross-level molecular interactions.

Purpose of the Study:

• To introduce MiDNE (Multi-omics genes and Drugs Network Embedding), a computational framework for integrating multi-omics data and pharmacological knowledge.
• To uncover disease-specific gene-drug interactions by analyzing complex molecular networks.
• To facilitate the discovery of novel therapeutic strategies through precision pharmacology.

Main Methods:

• MiDNE integrates omics-specific networks with drug-target interactions into a multiplex heterogeneous network.
• A random walk with restart algorithm is used for network embedding.
• Genes and drugs are projected into a shared latent space for clustering and neighborhood analysis.

Main Results:

• MiDNE successfully integrated multi-omics data (gene expression, methylation, proteomics, CNV) with drug-target interactions for Breast Invasive Carcinoma and Glioblastoma multiforme.
• The framework identified actionable gene-drug relationships by providing multilayer, disease-specific interaction views.
• Demonstrated potential for discovering precision pharmacological strategies.

Conclusions:

• MiDNE provides a powerful approach to model complex molecular interactions across multiple omics layers.
• The framework enables the identification of disease-specific gene-drug relationships for targeted therapies.
• MiDNE supports the advancement of precision medicine by uncovering novel pharmacological strategies.