iDRW

iDRW infers pathway activities from multi-omics genomic profiles by performing directed random walks on an integrated, multi-layered gene-gene graph constructed from pathway information to support molecular characterization and outcome prediction in cancer.

Key Features:

• Multi-layered Network Integration: Uses a directed gene-gene graph with interactions assigned across multiple network layers derived from pathway information.
• Pathway Activity Inference: Integrates multiple genomic profiles into unified pathway-level activity profiles via graph-based analysis.
• Directed Random Walks: Applies directed random walks on the integrated gene-gene graph to propagate signals and infer pathway activity.
• Outcome Prediction: Demonstrates improved outcome prediction performance in studies involving urologic cancer patients compared to single-profile methods.
• Identification of Prognostic Candidate Driver Pathways: Identifies common and cancer-specific candidate driver pathways and prioritizes genes within pathways as prognostic features.

Scientific Applications:

• Cancer pathway profiling: Characterizes molecular underpinnings and gene interactions within pathways using integrated multi-omics data.
• Prognosis and biomarker identification: Identifies prognostic pathways and candidate driver genes for outcome prediction, exemplified in urologic cancers.

Methodology:

Construct a directed gene-gene graph using pathway information with interactions across multiple network layers, then perform directed random walks on this graph to integrate genomic profiles into a cohesive pathway activity profile.