Signac

Signac performs analysis of single-cell chromatin and multimodal data, extending the Seurat toolkit to integrate DNA accessibility with gene expression, protein abundance, and mitochondrial genotype for tasks such as peak calling, clustering, and marker-gene identification.

Key Features:

• Chromatin data analysis: Implements peak calling, quantification, quality control, dimension reduction, clustering, and integration with single-cell gene expression data.
• Multimodal integration: Integrates chromatin (DNA accessibility) data with transcriptomic, proteomic, and mitochondrial-genotype modalities to correlate chromatin states with gene expression.
• Seurat extension: Extends the Seurat toolkit for multimodal single-cell analysis, enabling combined analyses within the Seurat framework.
• Scalability: Scales to large datasets, demonstrated on analyses involving over 700,000 cells.
• Marker-gene algorithm (Venice): Includes the Venice algorithm that defines marker genes by their ability to distinguish cell populations rather than by mean expression and avoids assumptions about specific distribution families.
• Gene-classification metric: Implements a metric to classify genes into up-regulated, down-regulated, and transitional states to improve marker identification.
• Benchmarking: Reports performance comparisons showing improvement over Seurat, ROTS, scDD, edgeR, MAST, limma, normal t-test, Wilcoxon, and Kolmogorov–Smirnov test for marker-gene identification.
• Performance optimization: Algorithmic implementation is optimized for computational speed and scalability for large single-cell datasets.

Scientific Applications:

• Single-cell epigenomics: Analysis of chromatin accessibility at single-cell resolution and its relationship to gene expression.
• Cellular heterogeneity studies: Identification of marker genes and chromatin features that distinguish cell populations within tissues or disease states.
• Integrative multi-omics analysis: Joint analysis of chromatin, transcriptomic, proteomic, and mitochondrial-genotype data for comprehensive single-cell multi-omics studies.

Methodology:

Computational steps explicitly include peak calling, quantification, quality control, dimension reduction, clustering, integration with single-cell gene expression data, the Venice marker-gene algorithm, a metric to classify genes as up-/down-regulated or transitional, and benchmarking against Seurat, ROTS, scDD, edgeR, MAST, limma, normal t-test, Wilcoxon, and Kolmogorov–Smirnov test.