Quartets enable statistically consistent estimation of cell lineage trees under an unbiased error and missingness...

Area of Science:

• Computational Biology
• Cancer Research
• Evolutionary Biology

Background:

• Reconstructing tumor evolutionary history aids cancer progression and treatment understanding.
• Traditional phylogenetic methods struggle with sparse, error-prone single-cell sequencing data and clonal evolution in tumors.
• Tumor phylogenetics requires robust methods that account for data imperfections.

Purpose of the Study:

• To investigate the theoretical utility of quartet-based phylogenetic methods for tumor evolution reconstruction.
• To address challenges posed by sparse data, errors, and unresolved evolutionary histories in tumor phylogenetics.
• To develop and validate a method for inferring unrooted cell lineage trees from mutation data.

Main Methods:

• Utilized a tumor phylogenetics model with mutations arising on an unresolved tree, followed by introduction of errors and missing values.
• Focused on quartets (four-leaf, unrooted phylogenetic trees) implied by mutations present in two cells and absent from two.
• Developed a method to find a tree maximizing shared quartets with input mutations, proving it as a consistent estimator.

Main Results:

• The most probable quartet accurately identifies the unrooted model tree for four cells.
• An optimal solution maximizing shared quartets consistently estimates the unrooted cell lineage tree.
• This estimation guarantee holds even for highly unresolved model trees and considers false negative branches as error.

Conclusions:

• Quartet-based phylogenetic methods offer a robust approach to reconstructing tumor evolutionary histories.
• These methods can overcome limitations of traditional phylogenetics when dealing with imperfect tumor sequencing data.
• Future work can extend quartet-based approaches to address complexities like copy number aberrations in tumor phylogenetics.