Human-scATAC-Corpus: a comprehensive database of scATAC-seq data.

Area of Science:

• Genomics
• Epigenetics
• Computational Biology

Background:

• Single-cell assay for transposase-accessible chromatin using sequencing (scATAC-seq) provides cellular resolution of chromatin accessibility.
• Heterogeneous data and complex pipelines hinder the creation of unified epigenomic resources.
• Existing resources lack comprehensive integration for diverse analytical needs.

Purpose of the Study:

• To establish a comprehensive, unified, and scalable database of human scATAC-seq data.
• To facilitate diverse downstream analyses, method development, and foundation model pretraining.
• To provide a standardized platform for evaluating single-cell epigenomic analyses.

Main Methods:

• Compilation of 5,407,621 cells from 35 scATAC-seq datasets across 37 human tissues/cell lines.
• Distribution of data in three aligned formats: cell-by-candidate cis-regulatory element matrices, raw fragment files, and cell-by-peak matrices.
• Integration of EpiAgent, a foundation model pretrained on the corpus for online analyses.

Main Results:

• Creation of Human-scATAC-Corpus, a large-scale database of human scATAC-seq data.
• Provision of data in multiple formats supporting cross-dataset integration, flexible processing, and within-dataset analysis.
• Demonstration of the resource's utility for method benchmarking, foundation model pretraining, and online analysis via EpiAgent.

Conclusions:

• Human-scATAC-Corpus provides a unified and scalable foundation for human single-cell epigenomics.
• The database accelerates discovery and standardizes evaluation in single-cell epigenomic research.
• The integrated EpiAgent enables advanced online analyses, enhancing resource accessibility and utility.