Scarf

Scarf performs memory-efficient preprocessing, analysis, and reference-anchored mapping of large-scale single-cell genomic datasets for single-cell RNA-Seq and ATAC-Seq studies.

Key Features:

• Modular Design: Provides interoperability with other single-cell toolkits to integrate analytical workflows.
• Memory Efficiency: Implements optimized, low-memory computations including graph-based t-stochastic neighbor embedding (t-SNE) and hierarchical clustering.
• Reference-Anchored Mapping: Maps query datasets to reference atlases while maintaining memory-efficient processing.
• Data Downsampling Algorithm: Uses a representative downsampling algorithm intended to preserve rare cell populations and lineage differentiation trajectories.
• Scalability on Standard Devices: Enables processing of atlas-scale single-cell datasets on commodity hardware such as laptops and low-cost single-board computers.

Scientific Applications:

• Single-Cell Transcriptomics: Analysis of gene expression profiles across millions of cells from single-cell RNA-Seq experiments.
• Epigenomic Profiling: Analysis of chromatin accessibility landscapes from single-cell ATAC-Seq data.
• Data Integration and Reanalysis: Integration and reanalysis of large datasets for comparative studies and meta-analyses against reference atlases.

Methodology:

Implements optimized graph-based t-SNE and hierarchical clustering, reference-anchored mapping to atlases, and a representative data downsampling algorithm.