scDEC-Hi-C

scDEC-Hi-C analyzes single cell Hi-C data using deep generative neural networks to impute sparse chromatin contact maps, cluster cells, and characterize 3D genome conformation variability at single-cell resolution.

Key Features:

• Deep Generative Modeling: Uses deep generative neural networks to model complex single cell Hi-C data distributions.
• Data Clustering and Imputation: Performs clustering of single cell Hi-C profiles and imputes missing contact data to reconstruct chromatin contact maps.
• Handling Sparse and Heterogeneous Data: Addresses sparsity and heterogeneity in single cell Hi-C datasets to improve analysis reliability.
• Chromatin Architecture Variability: Identifies cell-to-cell variability and differences in chromatin architecture across cell types.

Scientific Applications:

• Gene Regulation Studies: Supports studies linking 3D genome organization from single cell Hi-C to gene regulatory mechanisms and cellular differentiation.
• Cellular Function Analysis: Enables analysis of how cell-to-cell variability in chromatin organization relates to cellular functions and phenotypes.
• Comparative Genomics Across Cell Types: Facilitates comparison of chromatin architecture across cell types to investigate disease mechanisms and potential therapeutic targets.

Methodology:

Applies deep learning via deep generative neural networks to single cell Hi-C data for modeling, imputation, and clustering of sparse contact matrices.