UnpairReg

UnpairReg performs regression-based integration of unpaired single-cell multi-omics datasets to estimate gene expression from chromatin accessibility and infer cis-regulatory networks for improved cell-type characterization.

Key Features:

• Regression analysis on unpaired observations: Implements a novel regression approach tailored to integrate unpaired single-cell data across modalities.
• Integration of chromatin accessibility and gene expression: Jointly analyzes chromatin accessibility and RNA expression to link regulatory elements to genes.
• Estimation of gene expression from accessibility: Estimates cell gene expression profiles when only chromatin accessibility data are available.
• Cis-regulatory network inference: Infers cis-regulatory networks that are consistent with eQTL mapping results.
• Enhanced cell type identification: Improves cell-type classification through joint analysis of accessibility and expression data.

Scientific Applications:

• Cis-regulatory network discovery: Produces cis-regulatory maps for single cells that corroborate eQTL mapping findings.
• Cell type identification: Enhances accuracy of cell-type assignment in heterogeneous tissues by integrating modalities.
• Expression estimation from accessibility: Enables estimation of gene expression profiles from single-cell chromatin accessibility datasets.
• Method validation: Validated on both real and simulated datasets to assess performance on practical and controlled scenarios.

Methodology:

Applies a novel regression analysis for unpaired observations to jointly analyze chromatin accessibility and gene expression and to infer cis-regulatory networks.