Joint analysis of scATAC-seq datasets using epiConv.

Area of Science:

• Genomics
• Epigenetics
• Computational Biology

Background:

• Single-cell ATAC-seq (scATAC-seq) enables high-throughput chromatin state profiling.
• Technical variations, or batch effects, across scATAC-seq datasets hinder joint analysis.
• Specialized methods are needed to remove technical variations while retaining biological signals.

Purpose of the Study:

• To introduce epiConv, a novel algorithm for joint analysis of multiple scATAC-seq datasets.
• To demonstrate epiConv's ability to correct batch effects and preserve biological information.
• To improve the resolution and interpretability of scATAC-seq data.

Main Methods:

• Development of the epiConv algorithm for scATAC-seq data integration.
• Benchmarking epiConv against existing methods using PBMC and mouse brain datasets.
• Application of epiConv to integrate datasets from different biological conditions.

Main Results:

• epiConv demonstrates superior batch effect correction and reduced over-fitting compared to existing methods.
• The algorithm successfully aligns low-depth co-assay data to high-quality ATAC-seq references, enhancing chromatin profile resolution.
• Integration of diverse biological conditions revealed previously undetectable cell populations.

Conclusions:

• epiConv effectively integrates multiple scATAC-seq datasets, removing batch effects and preserving biological signals.
• Joint analysis using epiConv improves clustering and differentially accessible peak calling performance, especially for weak biological signals.
• The method facilitates deeper biological insights from scATAC-seq data.