NECo: A node embedding algorithm for multiplex heterogeneous networks.

Area of Science:

• Genomics
• Bioinformatics
• Systems Biology

Background:

• Complex diseases like hypertension, cancer, and diabetes are leading causes of death, involving intricate gene-environment interactions.
• Multi-omics datasets are increasingly available, driving the adoption of network-based methods for analyzing molecular interactions.
• Existing node embedding methods struggle to integrate diverse gene and phenotype data types.

Purpose of the Study:

• To develop a novel node embedding algorithm capable of integrating multilayered, heterogeneous networks of genes and phenotypes.
• To enhance the prediction of genes associated with complex diseases by leveraging integrated network analysis.

Main Methods:

• Developed Node Embeddings of Complex networks (NECo), a new node embedding algorithm.
• Utilized multilayered heterogeneous networks incorporating both genotypic and phenotypic data.
• Evaluated NECo performance using disease models from *Rattus norvegicus*.

Main Results:

• NECo significantly outperformed existing state-of-the-art node embedding methods in classifying hypertension-related genes.
• Achieved an Area Under the Curve (AUC) of 94.97%, compared to 85.98% for the second-best method.
• Identified novel genes potentially implicated in hypertension that were not previously recognized.

Conclusions:

• NECo offers a powerful approach for integrated network analysis of multi-omics data.
• The algorithm advances the identification of genetic factors contributing to complex diseases.
• NECo has the potential to improve our understanding and management of morbidity and mortality from complex diseases.