YusenYe/CellLoop: CellLoop

CellLoop identifies chromatin loops in individual cells using a density-based center detection algorithm, which integrates intra-cellular and neighboring inter-cellular contact maps through a re-voting strategy (Fig. 1a). By capturing both local topological associations and contextual support from neighboring cells, CellLoop enables robust identification of chromatin loops in sparse single-cell contact maps. Fig. 1b illustrates the analytical workflow and potential applications of CellLoop, including the visualization and comparison of LFmap and CFmap, the identification and visualization of cell-type-specific chromatin loops, the characterization of chromatin loop features that define distinct cell states, and linking cell-type-specific loops with gene expression. Several advantages include: (1). Loop Frequency Map (LFmap) obtained by CellLoop exhibited a significantly enhanced capacity. (2). CellLoop detected single-cell chromatin loops with an average time of merely 0.99s at 100kb resolution and ~13s at 10kb resolution in parallel with the support of 10 CPU processes, ensuring its capacity to process tens of thousands or even hundreds of thousands of single-cell data within a tolerable time. (3). CellLoop enables the signature of single cell from the perspective of chromatin loops. This, in turn, helps to facilitate the insight of subtle biological difference for single cells and further to reinterpret the biological functions of cell states or spatial domains from the perspective of 3D genomic chromatin loop-driven mechanisms.