Answers are grounded in this paper and cite their sources.
Uncovers improved clustering in single-cell hi-C, highlighting the role of structural-zero-aware measures.
High-throughput single-cell Hi-C technologies offer a powerful lens on cell-to-cell variability in the three-dimensional organization of the genome, yet their interpretation is constrained by extreme and uneven data sparsity. Zeros in contact maps can represent either genuine absence of contacts imposed by chromatin architecture (structural zeros) or missing observations due to insufficient sequencing depth (dropouts). However, current dissimilarity measures, including Euclidean distance and Kendall's tau, treat all zeros equivalently, obscuring biologically meaningful differences between cells. Here we introduce structural-zero-aware Kendall's tau (szKendall), an enhanced dissimilarity that leverages the spatial organization of two-dimensional contact maps and the concordance of shared structural zeros across cells. Through comprehensive simulations and analyses of real single-cell Hi-C datasets, we show improved capture of structural features and superior performance in cell clustering compared to existing approaches. Our results underscore the importance of structural-zero-aware dissimilarity measures as a principled foundation for robust inference from single-cell Hi-C data.
No takes yet. Share an insight, caveat, or question.
Liu et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: