SpaDC

SpaDC: Spatial Chromatin Accessibility Data Analysis

SpaDC is a graph-regularized convolutional neural network for spatial epigenomics. It jointly incorporates DNA sequence, chromatin accessibility, and spatial location to improve the analysis of spatial chromatin accessibility data.

Overview

SpaDC learns sequence-aware representations for denoising and integrating spatial epigenomic datasets. It supports spatial domain identification and regulatory inference, including the discovery of spatial domain-specific cis-regulatory elements and gene regulatory networks.

Key Features

• Sequence-aware modeling: uses DNA sequence together with chromatin accessibility signals
• Spatially informed learning: incorporates spatial-neighbor constraints for robust spatial domain identification
• Integrative analysis: aligns multiple spatial datasets and mitigates batch effects
• Regulatory inference: identifies cis-regulatory elements and infers transcription-factor gene regulatory networks

Publication & Code

This work was published in Communications Biology (2026).

• Read the paper
• View the code

(Ma et al., 2026)