pathifier

Pathifier infers pathway deregulation scores for individual tumor samples from gene expression data to provide pathway-level characterization of cancer-related biological processes.

Key Features:

• Per-sample pathway deregulation scores (PDS): Computes a pathway deregulation score for each individual tumor sample based on gene expression data.
• Gene-to-pathway transformation: Transforms gene-level expression information into a compact pathway-level representation.
• Dataset- and cancer-type specificity: Tailors pathway-level analysis to each specific dataset and cancer type.
• Deregulated pathway identification: Analyzes expression data to identify pathways that are deregulated within tumors.
• Preservation of gene expression integrity: Maintains the original gene expression information while summarizing at the pathway level.
• Validation on cancer datasets: Performance validated across multiple datasets, including three colorectal cancer and two glioblastoma multiforme datasets.
• Pathway-level survival markers: Identifies pathways linked to patient survival, including CXCR3-mediated signaling and oxidative phosphorylation in colorectal cancer.
• Tumor subclass discovery: Enables identification of tumor subclasses such as proneural and neural glioblastoma multiforme and an EGF receptor-deregulated subclass in colon cancer.

Scientific Applications:

• Survival association analysis: Detects biologically significant pathways associated with patient survival in glioblastoma multiforme and colorectal cancer.
• Tumor subclassification: Subclassifies tumors based on pathway deregulation patterns, distinguishing subtypes with different survival outcomes (e.g., proneural and neural GBM, EGF receptor-deregulated colon subclass).
• Pathway-level prognostic and therapeutic insight: Provides pathway-level insights that can inform prognostic assessments and potential targeted therapeutic strategies.

Methodology:

Analyzes gene expression data to compute per-sample pathway deregulation scores by transforming gene-level expression into pathway-level representations and identifying deregulated pathways, with validation reported on three colorectal cancer and two glioblastoma multiforme datasets.