Inferring clonal evolution of tumors from single nucleotide somatic mutations

Area of Science:

• Genomics
• Computational Biology
• Cancer Research

Background:

• High-throughput sequencing enables SNV frequency detection in tumor cells.
• SNV data can potentially reconstruct tumor cell evolutionary history and subclonal frequencies.
• Automated methods for this reconstruction are lacking, and conditions for success are undefined.

Purpose of the Study:

• To describe conditions for unique reconstruction of tumor evolutionary history from SNV frequencies.
• To introduce PhyloSub, a statistical model for inferring subclonal lineage phylogeny and genotypes.
• To address the lack of automated methods for reconstructing tumor evolution.

Main Methods:

• Developed PhyloSub, a Bayesian nonparametric model using priors over trees.
• PhyloSub groups SNVs into subclonal lineages and estimates lineage number and ancestry.
• Employs posterior distribution sampling to identify high-probability evolutionary histories and cell frequencies.

Main Results:

• PhyloSub uniquely reconstructs tumor evolutionary history under specified conditions.
• The model infers both linear and branching subclonal lineages.
• Experiments on simulated and real leukemia data show good agreement with ground truth.

Conclusions:

• PhyloSub is applicable to binary somatic mutations, including SNVs, insertions, and deletions.
• The software is available for broader research application.
• Enables robust inference of tumor phylogeny and subclonal architecture.