ConDoR: Tumor phylogeny inference with a copy-number constrained mutation loss model

Area of Science:

• Computational Biology
• Genomics
• Cancer Research

Background:

• Tumors comprise diverse cell subpopulations with unique somatic mutations, including single nucleotide variants (SNVs) and copy-number aberrations (CNAs).
• Accurate reconstruction of tumor phylogenies is crucial for understanding cancer evolution, but current single-cell DNA sequencing (scDNA-seq) technologies struggle to reliably measure both SNVs and CNAs simultaneously.
• Existing phylogenetic inference methods can be misled by CNAs that overlap SNVs, potentially causing errors in evolutionary models.

Approach:

• Introduced the constrained k-Dollo model, an evolutionary model that leverages SNVs as phylogenetic markers and incorporates partial CNA information through cell clustering based on copy-number profiles.
• Developed ConDoR (Constrained Dollo Reconstruction), an algorithm designed to infer tumor phylogenies from targeted scDNA-seq data using the constrained k-Dollo model.
• Validated ConDoR's performance against existing methods using simulated data, demonstrating superior accuracy.

Key Points:

• ConDoR effectively constrains SNV loss events within the phylogenetic tree by utilizing copy-number clustering.
• Applied ConDoR to a multi-region targeted scDNA-seq dataset from a pancreatic ductal adenocarcinoma (PDAC) tumor (2153 cells), yielding a more plausible tumor phylogeny consistent with histological findings.
• Analyzed a metastatic colorectal cancer dataset, generating a more parsimonious phylogeny with a simpler monoclonal origin of metastasis compared to previous studies.

Conclusions:

• The constrained k-Dollo model and ConDoR algorithm offer a significant advancement in reconstructing accurate tumor phylogenies from targeted scDNA-seq data.
• ConDoR provides more reliable evolutionary insights into cancer, particularly for complex tumor types like PDAC and metastatic colorectal cancer.
• This approach enhances our understanding of tumor heterogeneity and evolutionary trajectories, aiding in the development of targeted cancer therapies.