CLAM

CLAM integrates multi-omics data and known molecular interactions to detect co-regulated gene modules for biomarker discovery and biological interpretation.

Key Features:

• Trans-Omics Neighborhood Matrix Construction: CLAM constructs a trans-omics neighborhood matrix by integrating diverse multi-omics datasets with transcriptional regulatory interactions, protein-protein interactions, and biological pathways.
• Local Approximation Procedure: CLAM employs a local approximation procedure to define and detect co-regulated gene modules from the neighborhood matrix.
• Integration Across Incomplete Datasets: CLAM overcomes the requirement for genes or samples to be present in every dataset by leveraging known molecular interactions during integration.
• Module-Based Survival Analysis: CLAM supports module-based survival analysis to identify networks whose gene correlations significantly correlate with patient survival.
• Functional Characterization: CLAM facilitates recovery of gene ontology (GO) terms and identification of key transcription factors and KEGG pathways associated with detected modules.

Scientific Applications:

• Modular biomarker discovery in colorectal cancer (CRC): CLAM identified transcription factors and KEGG pathways involved in CRC progression through module detection.
• Cross-species module analysis: CLAM was applied to mouse B-cell differentiation datasets to detect co-regulated gene modules.
• Data integration across major repositories: CLAM has been evaluated using datasets from The Cancer Genome Atlas (TCGA), Clinical Proteomics Tumor Analysis Consortium (CPTAC), Gene Expression Omnibus (GEO), and the ProteomeXchange database.
• Prognostic network identification: CLAM enabled identification of networks where module correlations correlate with patient survival in module-based analyses.

Methodology:

CLAM constructs a trans-omics neighborhood matrix by integrating multi-omics datasets with transcriptional regulatory interactions, protein-protein interactions, and biological pathways, then applies a local approximation procedure to define and detect gene modules.