Identifying tumor clones in sparse single-cell mutation data.

Area of Science:

• Genomics
• Computational Biology
• Cancer Research

Background:

• Single-cell DNA sequencing offers whole-genome insights but suffers from ultra-low coverage (<0.5× per cell).
• This sparsity limits analysis to large copy-number aberrations (CNAs), hindering the study of single-nucleotide mutations crucial for cancer research.
• Existing methods struggle with sparse single-nucleotide mutation data from single cells.

Purpose of the Study:

• To introduce SBMClone, a novel computational method for inferring clonal structures from sparse single-nucleotide mutation data.
• To demonstrate SBMClone's accuracy in identifying cell clusters (clones) even with ultra-low sequencing coverage.
• To apply SBMClone to real-world single-cell whole-genome sequencing data from breast cancer patients.

Main Methods:

• Developed SBMClone, a method employing a stochastic block model to address data sparsity.
• Validated SBMClone using simulated datasets with sequencing coverage as low as 0.2×.
• Applied SBMClone to single-cell whole-genome sequencing data from two breast cancer patients using different technologies (10X Genomics CNV and DOP-PCR).

Main Results:

• SBMClone accurately inferred clonal composition on simulated data, demonstrating robustness at low coverage.
• For a breast cancer patient (≈0.03× coverage), SBMClone recovered major clonal structures with minimal additional information.
• Analysis of a second patient's data (≈0.5× coverage) revealed the presence of tumor cells post-treatment, contradicting previous findings.

Conclusions:

• SBMClone effectively overcomes the sparsity challenge in ultra-low coverage single-cell sequencing for single-nucleotide mutation analysis.
• The method accurately reconstructs clonal architecture and provides valuable insights into cancer heterogeneity and evolution.
• SBMClone offers a powerful tool for advancing single-cell genomics in cancer and other research areas.