BifurcatoR

BifurcatoR detects, models, and interprets variance heterogeneity (VH) in biological data through simulation-based method evaluation and study design recommendations.

Key Features:

• Variance Heterogeneity Detection: Identifies VH patterns including skewness, outliers, and multimodal distributions that may signal biologically distinct subgroups.
• Simulation-Based Method Evaluation: Benchmarks VH methods through simulation to assess performance across different data distributions.
• Study Design Recommendations: Provides guidance on sample size and method selection for VH analysis.
• Method Integration: Integrates multiple VH methods to support analysis across a range of data distributions, including those that violate normality assumptions.

Scientific Applications:

• Patient Stratification: Identifies latent etiological structures and molecularly distinct subgroups with different treatment responses, as demonstrated in acute myeloid leukemia (AML) where an EVI1-high group with significantly poorer prognosis was identified.
• Biomarker Discovery: Reveals clinically actionable signals in variance patterns relevant to prognosis and therapeutic response.
• Precision Medicine: Supports incorporation of VH into biomedical research for understanding disease heterogeneity and treatment outcomes, as applied in obesity and AML clinical datasets.

Methodology:

BifurcatoR benchmarks VH methods through simulation studies evaluating performance across distribution shape, mean-variance coupling, and underlying subgroup structure.