dualmarker

dualmarker evaluates and visualizes combinatorial dual biomarkers to assess and compare their ability to predict therapy response and patient survival.

Key Features:

• Data exploration and hypothesis generation: Supports comprehensive exploration of clinical biomarker datasets to generate hypotheses about dual-marker interactions.
• Visualizations (14 plot types): Produces 14 plot types including boxplots, scatterplots, ROC curves, and Kaplan-Meier plots to display associations between two biomarkers and drug response or survival.
• Statistical modeling: Implements logistic regression for binary outcomes and Cox regression for time-to-event outcomes, and compares model fitness to assess whether dual markers outperform single markers.
• Workflow streamlining: Streamlines the analysis flow to support de novo searches for novel biomarker combinations.

Scientific Applications:

• Immune checkpoint blockade trials: Evaluate and validate combinatorial biomarkers for predicting response to immune checkpoint inhibitors in clinical cohorts.
• Patient stratification and prognosis: Identify biomarker pairs that improve prediction of therapy response and overall survival for personalized medicine applications.

Methodology:

Uses logistic regression for binary response analysis and Cox regression for survival analysis, comparing model fitness between dual-marker and single-marker models; demonstrated on the IMvigor210 bladder cancer dataset with examples including TMB/TGF-beta signature for response, CXCL13 expression/ARID1A mutation for survival, and HMGB1 expression/ARID1A mutation for overall survival.