EpiCarousel

EpiCarousel identifies metacells in large-scale single-cell chromatin accessibility data through memory- and time-efficient computational strategies.

Key Features:

• Lazy Loading: Enables memory-efficient processing of atlas-level datasets by loading data only when needed.
• Parallel Processing: Distributes computational workload to reduce execution time for large-scale analyses.
• Community Detection: Identifies metacells representing homogeneous cell groups in single-cell chromatin accessibility sequencing data.
• Preprocessing Integration: Executes preprocessing steps as part of the metacell identification workflow.
• Downstream Cell Clustering: Performs cell clustering on atlas-level datasets as part of the analysis pipeline.

Scientific Applications:

• Atlas-Scale scCAS Analysis: Processes datasets with hundreds of thousands of cells and over one million peaks, such as 707 043 cells and 1 154 611 peaks.
• Cell Type Identification: Supports downstream analysis for identifying cell types from metacell representations.
• Characterization of Epigenetic Heterogeneity: Enables analysis of cell-to-cell variation in chromatin accessibility patterns across diverse protocols, sample sizes, dimensions, number of cell types, and degrees of cell-type imbalance.

Methodology:

EpiCarousel employs lazy loading, parallel processing, and community detection methods for metacell identification in single-cell chromatin accessibility sequencing data.