BRANEnet: embedding multilayer networks for omics data integration.

Area of Science:

• Systems Biology
• Bioinformatics
• Genomics

Background:

• Gene expression is regulated across multiple molecular levels, intricately linked with cellular metabolism.
• Integrating multi-omics data is crucial for understanding complex cellular systems.
• Existing methods face challenges in handling multilayer heterogeneous networks.

Purpose of the Study:

• To introduce BRANENET, a novel framework for multi-omics data integration.
• To enable the study of regulatory aspects in multilayered biological processes.
• To provide a scalable and versatile method for learning node embeddings.

Main Methods:

• BRANENET utilizes a matrix factorization framework with random walk information.
• The framework is designed for multilayer heterogeneous networks.
• It learns node embeddings to represent biological entities.

Main Results:

• BRANENET was evaluated using transcriptomics (RNA-seq) and metabolomics (NMR) data from Saccharomyces cerevisiae.
• The framework successfully identified features of differentially expressed biomolecules during heat stress.
• Learned features improved performance in transcription factor-target prediction, integrated omics network inference, and module identification.

Conclusions:

• BRANENET offers an effective approach for multi-omics data integration.
• The framework demonstrates superior performance compared to existing network integration methods.
• BRANENET facilitates deeper insights into cellular regulatory mechanisms.