NetRank

NetRank identifies biomarker signatures for cancer prediction by integrating RNA-seq and multi-omics data with network-based algorithms to prioritize genes associated with phenotypic outcomes.

Key Features:

• Integration of Multi-Omics Data: Combines various omics data types, including RNA-seq gene expression profiles, to inform biomarker selection for disease progression and treatment response prediction.
• Network-Based Algorithm: Constructs protein-protein interaction networks using StringDB and builds weighted gene co-expression networks with WGCNA to capture molecular interactions.
• Network-Informed Prioritization: Leverages protein associations, co-expression patterns, and functional relationships with phenotypic outcomes to rank candidate biomarkers.
• Biomarker Discovery: Identifies interpretable biomarker signatures that differentiate between cancer types using RNA-seq gene expression data.
• Robust Feature Selection: Assesses robustness and suitability of RNA gene expression features across 19 cancer types using genomic data from over 3,000 TCGA patients.
• High Predictive Performance: Produces compact biomarker signatures that segregate most cancer types with area under the curve (AUC) values reported above 90%.
• Implementation Components: Includes pre- and post-processing functions for RNA-seq gene expression data and functions for building protein-protein interaction and co-expression networks.

Scientific Applications:

• Oncology Biomarker Discovery: Prioritizes molecular signatures associated with distinct cancer types for diagnostic and prognostic investigations.
• Cancer Outcome Prediction: Generates gene signatures from RNA-seq and network features to predict disease progression and treatment response.
• Pan-Cancer Evaluation: Facilitates comparative assessment of biomarker robustness across 19 TCGA cancer types and cohorts totaling over 3,000 patients.

Methodology:

Uses RNA-seq gene expression data and TCGA genomic datasets to construct protein-protein interaction networks via StringDB and weighted gene co-expression networks via WGCNA, and applies network-based ranking algorithms to prioritize interpretable biomarkers based on protein associations, co-expression, and functional relationships with phenotypic outcomes.