SCArray.sat

SCArray.sat extends Seurat classes and functions to support Genomic Data Structure (GDS) files as a DelayedArray backend for single-cell RNA-seq data representation and analysis.

Key Features:

• GDS-based DelayedArray backend: Relies on the SCArray package implementation of GDS-based DelayedMatrix to represent single-cell RNA-seq data.
• SC_GDSMatrix optimization: Leverages common optimized algorithms for GDS-based and single-cell-specific DelayedMatrix implemented in the SCArray package.
• SCArrayAssay class: Introduces a new class derived from Seurat Assay that wraps raw counts, normalized expressions, and scaled data matrix based on GDS-specific DelayedMatrix.
• Seurat integration: Integrates seamlessly with the Seurat package to provide common data analysis in the SeuratObject-based workflow.
• Memory efficiency: Significantly reduces memory usage without downsampling compared with standard Seurat, enabling application to very large datasets.

Scientific Applications:

• Large-scale single-cell RNA-seq analysis: Enables analysis of very large single-cell RNA-seq datasets that exceed available memory with standard Seurat workflows.
• Seurat-compatible workflows: Supports common single-cell data analysis within the established SeuratObject-based framework without requiring workflow changes.

Methodology:

SCArray.sat implements a GDS-specific DelayedMatrix backend for data representation and introduces the SCArrayAssay class derived from Seurat Assay to wrap raw counts, normalized expressions, and scaled data matrices.