PhySigs

PhySigs infers clone-level mutational signature exposures within a phylogenetic framework to analyze evolutionary dynamics of tumor clones.

Key Features:

• Input Requirements: Accepts a phylogeny of tumor clones, a signature matrix of mutational signatures, and a feature matrix of mutations per clone with categories aligned to the signature matrix.
• Count Matrix Computation: Computes a count matrix as a diagonal matrix derived from summing columns of the feature matrix.
• Tree-constrained Exposure (TE) problem: Solves the TE problem by evaluating all clone clusterings that correspond to partitions of the phylogeny and outputs a relative exposure matrix in which identical columns indicate clusters of clones with similar exposures.
• Identification of Exposure Shifts: Identifies edges between clusters in the phylogeny that indicate shifts in mutational signature exposures.
• Model Selection and Validation: Performs model selection to determine the optimal number of exposure shifts and validates results using simulations and real-world datasets such as lung cancer.
• Phylogeny Prioritization: Facilitates prioritization of alternative phylogenies derived from the same sequencing data.

Scientific Applications:

• Cancer evolutionary analysis: Analyze evolutionary trajectories of tumors by integrating mutational signatures with clone phylogenies.
• Detection of mutational process shifts: Detect and localize shifts in mutational signature exposures and relate them to subclonal driver mutations and pathways such as mismatch repair in datasets including lung cancer.
• Phylogeny evaluation: Prioritize alternative phylogenetic reconstructions derived from the same sequencing data to improve evolutionary inference.

Methodology:

Requires as input a clone phylogeny, a signature matrix, and a feature matrix; computes a diagonal count matrix by summing feature-matrix columns; solves the Tree-constrained Exposure problem by evaluating all phylogeny partitions to produce a relative exposure matrix; performs model selection for the number of exposure shifts and validates findings via simulations and application to lung cancer datasets; facilitates prioritization of alternative phylogenies.