Heterogeneous Multi-Layered Network Model for Omics Data Integration and Analysis.

Area of Science:

• Computational Biology
• Bioinformatics
• Systems Biology

Background:

• Omics data generation is rapidly advancing, creating vast, complex datasets.
• Integrating and analyzing diverse omics data presents significant computational challenges.
• Biological networks are crucial for representing and inferring relationships between biological entities.

Purpose of the Study:

• To review recent computational methods for inferring novel biological relations from Heterogeneous Multi-Layered Networks (HMLNs).
• To discuss the properties and applications of HMLNs in omics data integration.
• To highlight future directions in HMLN model development for biological discovery.

Main Methods:

• Survey of four categories of state-of-the-art computational methods: matrix factorization, random walk, knowledge graph, and deep learning.
• Discussion of the unique properties of biological HMLNs.
• Demonstration of HMLN applications in omics data integration and analysis.

Main Results:

• HMLNs effectively integrate diverse biological data, representing biological system hierarchies.
• Various computational approaches (matrix factorization, random walk, knowledge graphs, deep learning) are applicable to HMLN analysis.
• These methods facilitate the inference of novel biological relations, aiding in understanding genotype-phenotype associations and environmental impacts.

Conclusions:

• Novel computational methods are advancing the inference of biological relations from HMLNs.
• HMLNs offer powerful frameworks for integrating and analyzing complex omics data.
• Future research should focus on developing new HMLN models to address remaining computational challenges.