Identification of functional gene modules by integrating multi-omics data and known molecular interactions

Area of Science:

• Computational biology
• Bioinformatics
• Systems biology

Background:

• Multi-omics data integration is crucial for identifying patient subgroups.
• Existing methods have limitations in handling missing data and incorporating molecular interactions.
• Gene co-expression module detection is a key challenge in multi-omics analysis.

Purpose of the Study:

• To present a novel data integration framework, Correlation-based Local Approximation of Membership (CLAM).
• To address limitations of existing methods by integrating multi-omics data with molecular interactions.
• To identify biologically relevant gene modules and potential biomarkers for complex diseases.

Main Methods:

• Developed a framework, Correlation-based Local Approximation of Membership (CLAM).
• Constructed a trans-omics neighborhood matrix integrating multi-omics datasets and molecular interactions.
• Employed a local approximation procedure for gene module definition.

Main Results:

• CLAM demonstrated superior ability in recovering biologically relevant modules and gene ontology (GO) terms.
• Applied CLAM to colorectal cancer (CRC) and mouse B-cell differentiation data.
• Identified key transcription factors and KEGG pathways in CRC progression and constructed survival-related networks.

Conclusions:

• Correlation-based Local Approximation of Membership (CLAM) shows great potential for identifying modular biomarkers.
• The framework effectively integrates multi-omics data and molecular interactions for improved module detection.
• CLAM facilitates the discovery of novel insights into complex disease mechanisms and patient stratification.