CNVeil is a haplotype-aware analytical framework designed to infer total, allele-specific, and chromosome-scale haplotype-resolved copy number from single-cell DNA sequencing (scDNA-seq) data. The tool addresses challenges specific to scDNA-seq — low coverage, allelic dropout, and technical noise — that make phased allelic inference substantially harder than total copy number estimation. The authors position CNVeil to close a methodological gap by converting sparse allelic signal across single cells into coherent, chromosome-scale haplotype profiles to study tumor heterogeneity and clonal evolution.
CNVeil begins by building a robust total copy number profile that serves as a stable scaffold for subsequent allelic and haplotype inference. Key elements of this stage include:
This scaffold is intended to stabilize downstream inference by providing consistent segment boundaries and baseline total-copy states despite low per-cell coverage.
Using the consensus total copy number profile as a scaffold, CNVeil infers allele-specific copy number by modeling heterozygous SNP allele counts. The framework applies an expectation–maximization (EM) algorithm to allelic read counts at heterozygous sites across cells and segments. The EM approach estimates the allele-specific copy number configuration that best explains observed allelic proportions while leveraging the stability of the total-copy scaffold to mitigate the impact of allelic dropout and sparse counts.
After allele-specific states are estimated per segment, CNVeil reconstructs haplotype-specific copy number by enforcing a coherent haplotype orientation across adjacent segments. This step uses dynamic programming to find a consistent assignment of haplotype labels along chromosomes, producing chromosome-scale haplotype-resolved copy number profiles for individual cells. The dynamic programming enforcement reduces phase-switching errors and yields contiguous haplotype calls across segments.
The authors benchmarked CNVeil against 12 state-of-the-art methods: eight total copy number callers, two allele-specific callers, and two haplotype-resolved callers. Benchmarks used 20 simulated and real datasets covering six experimental settings, including high-multiplexed single-nucleus sequencing, Acoustic Cell Tagmentation (ACT), and 10x Chromium. According to the report, CNVeil consistently outperformed existing tools on metrics that include segmentation accuracy, ploidy inference, subclone identification, and allele-specific copy number estimation. This comparison is described as the largest comparative evaluation of single-cell copy number inference methods to date.
Applied to a breast cancer multi-omics cohort generated with the wellDR-seq protocol, CNVeil revealed haplotype-specific subclonal diversification that was not apparent from total copy number analysis alone. In this cohort, the authors linked allele-specific copy number states to transcriptional variation, suggesting that haplotype-resolved copy number can expose genotype–phenotype relationships at single-cell resolution that total copy number profiles miss.
By transforming sparse allelic signals into chromosome-scale haplotype-resolved profiles, CNVeil aims to enable more precise characterization of tumor subclones, allelic imbalance, and clonal evolutionary trajectories. Haplotype-resolved copy number can reveal subclonal events masked in total copy number profiles, help associate allelic states with gene expression, and provide a richer substrate for studies of tumor biology and potential therapeutic resistance mechanisms.
This work is presented as a preprint and has not been certified by peer review. The abstract references benchmarking and application to multiple experimental settings and a breast cancer multi-omics cohort, but the abstract does not provide exhaustive methodological parameters, runtime characteristics, or specific availability details for code and data in this summary. For full technical details, supplementary materials, and any software release notes, readers should consult the full preprint and its supplementary information.