Hintra

Hintra infers tumor phylogenies and deconvolutes phylogenetic relationships among cancer mutations to detect intra-tumor heterogeneity from bulk sequencing data.

Key Features:

• Phylogenetic deconvolution: Deconvolutes phylogenetic relationships between cancer mutations to resolve subclonal structure.
• Cohort-level integration: Integrates sequencing data across a cohort of tumors rather than analyzing individual samples in isolation.
• Shared evolutionary information: Leverages evolutionary information shared between different tumors within the same cohort to improve inference.
• Mutation ordering: Infers phylogenetic orders of mutations at the individual-sample level.
• Iterative pattern learning: Employs an iterative process to identify and learn repeating evolutionary patterns across tumor samples in a cohort.
• Ambiguity resolution: Resolves phylogenetic ambiguities that commonly hinder traditional methods.
• Bulk sequencing applicability: Operates on bulk sequencing data obtained from single tumor samples.
• Robustness to noise: Addresses noise in sequencing data and complexity of biological mechanisms.
• Empirical benchmarking: Demonstrated superior performance in synthetic experiments compared to two state-of-the-art methods.
• Breast Cancer application: Produced results on a recent Breast Cancer dataset that align with established knowledge and indicate potentially novel findings.

Scientific Applications:

• Intra-tumor heterogeneity detection: Detects and characterizes subclonal heterogeneity within tumors from bulk sequencing data.
• Tumor phylogeny reconstruction: Reconstructs tumor phylogenies and orders mutations within individual samples.
• Ambiguity mitigation in bulk data: Resolves phylogenetic ambiguities arising from single-sample bulk sequencing.
• Cohort-level evolutionary analysis: Identifies repeating evolutionary patterns shared across a cohort of tumors.
• Method benchmarking: Provides a basis for comparison against existing state-of-the-art methods using synthetic experiments.
• Cancer dataset analysis: Applied to Breast Cancer datasets to reproduce known patterns and suggest novel hypotheses.

Methodology:

Integrates sequencing data across a tumor cohort and iteratively identifies and learns repeating evolutionary patterns to deconvolute phylogenetic relationships and infer mutation order for individual samples.