InferLoop

InferLoop infers chromatin interaction strengths from single-cell chromatin accessibility data to characterize cell-type-specific three-dimensional chromatin structures.

Key Features:

• Cell binning for signal enhancement: Groups nearby cells into bins to enhance chromatin accessibility signal for downstream analysis.
• Pearson perturbation-like metric: Infers loop signals from accessibility using a novel metric akin to the perturbation of the Pearson correlation coefficient.
• Single-cell resolution: Enables inference of chromatin loops at single-cell resolution.
• Gene expression prediction: Links inferred loop signals to the prediction of gene expression levels.
• Intergenic loci interpretation: Integrates GWAS Catalog and GTEx information to interpret intergenic loci in the context of inferred loops.
• Compatibility with diverse data types: Validated on and applicable to single-cell 3D genome structure data, single-cell multi-omics data, spatial chromatin accessibility data (including mouse embryo), and datasets from human brain cortex, human blood, and mouse brain cortex.

Scientific Applications:

• Inference of cell-type-specific loop signals: Identifies and characterizes chromatin loops that are specific to distinct cell types to reveal regulatory landscapes.
• Prediction of gene expression levels: Uses inferred loop signals to aid prediction of transcriptional activity and link chromatin architecture to gene expression.
• Interpretation of intergenic loci: Associates inferred loops with GWAS Catalog and GTEx annotations to provide insight into non-coding regulatory roles and disease mechanisms.
• Spatial loop prediction: Predicts loop signals at individual spots in spatial chromatin accessibility datasets such as the mouse embryo.

Methodology:

Groups nearby cells into bins to enhance accessibility signal, then leverages accessibility signals to infer loop signals using a novel metric akin to the perturbation of the Pearson correlation coefficient.