GRMT

GRMT reconstructs tumor mutation trees from single-cell DNA sequencing (SCS) data to recover the chronological order of mutations and infer tumor evolutionary histories.

Key Features:

• k-Dollo parsimony: Implements the k-Dollo parsimony model in which each mutation can be gained once and lost up to k times.
• Iterative de novo tree construction: Generates tree structures from scratch and incrementally increases tree size by introducing new mutations until all detected mutations are included.
• Bayesian optimization (BayesOpt) for error estimation: Uses Bayesian optimization to estimate error rates in mutation data.
• Noise-aware analysis of SCS data: Explicitly addresses the inherent noise in single-cell DNA sequencing data during tree reconstruction.
• Chronological mutation recovery: Recovers the chronological order of mutations in the inferred tumor phylogeny.
• Scalability: Designed to scale to large single-cell datasets.
• Empirical validation: Evaluated on simulated and real datasets and reported to outperform state-of-the-art methods across multiple performance metrics.

Scientific Applications:

• Tumor phylogeny reconstruction: Inferring mutation trees to study tumor evolutionary histories from SCS data.
• Clonal architecture inference: Determining clonal composition and relationships within tumors.
• Mutation chronology estimation: Ordering somatic mutations to elucidate the temporal sequence of tumor evolution.
• Method benchmarking: Comparing and validating phylogeny reconstruction approaches on simulated and real SCS datasets.

Methodology:

Uses the k-Dollo parsimony model, an iterative de novo tree-construction process that grows trees by adding mutations until all detected mutations are included, and Bayesian optimization (BayesOpt) to estimate error rates on single-cell DNA sequencing (SCS) data.