A Novel Kernel-Based Hilbert Space Framework for Predictive Modeling of lncRNA-miRNA-Disease Interaction Networks

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Long non-coding RNAs (lncRNAs) and microRNAs (miRNAs) are key regulators of gene expression implicated in diseases like cancer and neurodegenerative disorders.
• lncRNA-miRNA-disease networks (LMDNets) are crucial for understanding disease mechanisms but existing computational models face challenges in interpretability, scalability, and data noise.

Purpose of the Study:

• To develop a robust and interpretable computational framework, KHSF-LMDNet, for modeling lncRNA-miRNA-disease networks.
• To overcome limitations of existing methods, including poor interpretability, reliance on manual curation, and sensitivity to noisy or missing data.

Main Methods:

• Proposed KHSF-LMDNet, a kernel-based Hilbert space framework integrating graph-based networks, similarity features, and deep learning with an attention mechanism.
• Mapped complex lncRNA-miRNA-disease interactions into a Hilbert subspace for enhanced learning.
• Evaluated performance on benchmark datasets using accuracy, precision, and AUC metrics.

Main Results:

• KHSF-LMDNet demonstrated superior performance compared to existing methods in accuracy, precision, and AUC.
• The model effectively ranked disease-associated lncRNAs and miRNAs.
• Identified top candidate lncRNAs and miRNAs linked to cancer and Alzheimer's disease.

Conclusions:

• KHSF-LMDNet offers a more robust and interpretable approach to modeling LMDNets.
• The framework supports functional genomics research and facilitates the discovery of novel biomarkers for precision medicine.
• Highlights the potential of advanced computational methods in understanding complex disease-associated gene regulatory networks.