scGAD

scGAD computes gene associating domain (scGAD) scores that summarize gene-level chromatin contact patterns from single-cell Hi-C (scHi-C) data to enable dimensionality reduction, bias-corrected representation, and integration with other single-cell modalities.

Key Features:

• Gene-level summarization: Computes scGAD scores that summarize the association of genes with chromatin domains at single-cell resolution.
• Dimension reduction: Produces low-dimensional projections for exploratory analysis and visualization of scHi-C data.
• Accounting for genomic biases: Adjusts for inherent gene-level genomic biases to improve representation accuracy.
• Clustering by 3D structure: Captures clustering of cells driven by three-dimensional chromatin structures in low-dimensional spaces.
• Identification of significant interactions: Identifies genes with significant chromatin interactions within and across cell types.
• Integration with other modalities: Enables projection of scHi-C data onto reference low-dimensional embeddings to facilitate integration with other single-cell datasets.

Scientific Applications:

• 3D genome organization: Characterizing single-cell three-dimensional chromatin architecture and its variation across cells.
• Developmental biology: Investigating chromatin architecture changes during cellular differentiation.
• Cancer research: Profiling altered chromatin interactions and cell-type heterogeneity in cancer.
• Epigenetics: Linking chromatin architecture to gene regulation and epigenetic states.

Methodology:

Computes scGAD scores that summarize gene–chromatin domain associations at single-cell resolution, accounts for gene-level genomic biases, generates low-dimensional projections for clustering, enables projection onto reference low-dimensional embeddings, and identifies genes with significant chromatin interactions.