SBMClone

SBMClone infers clonal structures from sparse single-cell DNA sequencing data by applying stochastic block model (SBM) inference to somatic single-nucleotide mutation profiles, enabling clonal composition analysis at ultra-low coverage.

Key Features:

• Stochastic Block Model Inference: SBMClone employs stochastic block model (SBM) inference to cluster cells that share somatic single-nucleotide mutations and to infer clonal compositions.
• Ultra-Low Coverage Support: The method is tailored for ultra-low coverage single-cell sequencing data, with demonstrated performance on datasets down to ~0.2× per cell.
• Mutation Type Versatility: SBMClone applies to single-nucleotide variants and can be extended to structural variants and large copy-number aberrations (CNAs).
• Validation on Simulated and Real Data: The approach has been validated on simulated datasets and applied to single-cell whole-genome sequencing data from breast cancer patients to recover clonal compositions.
• Case Studies with Sequencing Technologies: SBMClone recovered major clonal compositions from a 10X Genomics CNV dataset at ~0.03× coverage and identified tumor cells in pre/post-treatment samples sequenced by DOP-PCR at ~0.5× coverage.

Scientific Applications:

• Cancer clonal evolution: Infer clonal lineages and mutation sharing among cells to study tumor evolution using somatic single-nucleotide variants.
• Tumor heterogeneity and treatment monitoring: Detect residual tumor cells and compare pre- and post-treatment clonal compositions from single-cell whole-genome sequencing datasets.
• Analysis of ultra-low coverage datasets: Enable mutation-level analyses in datasets generated by technologies such as the 10X Genomics CNV solution and DOP-PCR at very low coverage.

Methodology:

SBMClone performs stochastic block model (SBM) inference to cluster cells by shared somatic single-nucleotide variants and infer clonal compositions from ultra-low coverage single-cell DNA sequencing data.