Reconstruction of clonal trees and tumor composition from multi-sample sequencing data.

Area of Science:

• Computational Biology
• Genomics
• Cancer Research

Background:

• Tumorigenesis involves the evolution of distinct cell populations.
• Analyzing tumor heterogeneity requires understanding clonal architecture.

Purpose of the Study:

• To develop a computational method for reconstructing tumor clonal evolution.
• To address the challenge of inferring mutation relationships from sequencing data.

Main Methods:

• Formalized tumor evolution reconstruction as a variant allele frequency (VAF) factorization problem.
• Developed an integer linear programming approach for VAF factorization.
• Extended the method to handle sequencing errors using a probabilistic model.
• Implemented the AncesTree algorithm for analyzing mutation ancestry.

Main Results:

• Demonstrated that the VAF factorization problem is NP-complete.
• The AncesTree algorithm accurately identifies ancestral relationships between mutations.
• AncesTree outperforms existing methods, particularly with high-confidence VAFs from ultra-deep sequencing.

Conclusions:

• AncesTree provides a robust computational framework for tumor clonal evolution analysis.
• The VAF factorization approach is effective for reconstructing complex tumor phylogenies.
• Accurate reconstruction of clonal architecture is crucial for understanding cancer development.