TL;DR
Researchers have developed ATACdb, a comprehensive database that aggregates and annotates human chromatin accessibility data from ATAC-seq experiments. This resource aims to make this crucial information more accessible for studying gene regulation, disease mechanisms, and cell-specific functions by providing standardized data, quality control metrics, and integrated annotations.
What Problem Does This Paper Address?
Understanding chromatin accessibility is vital for deciphering gene regulation, cell differentiation, and the development of diseases. ATAC-seq is a powerful technique for mapping these accessible regions, but the rapid accumulation of ATAC-seq datasets has outpaced the development of centralized, well-annotated resources. Existing databases often focus on other epigenomic data types (like DNase-seq) or have limited scope in terms of sample size, annotation depth, or specific analysis tools for ATAC-seq data. This creates a need for a comprehensive, user-friendly platform that not only collects a large volume of human ATAC-seq data but also provides robust quality control, detailed annotations, and analytical tools to facilitate research.
Methods
The ATACdb database was constructed through a multi-step process:
- Data Collection: Over 2200 human ATAC-seq samples were manually curated from NCBI GEO/SRA. Non-compliant samples, such as single-cell ATAC-seq, were filtered out.
- Read Processing: Raw sequencing reads were trimmed for adapters and low-quality sequences using Trim Galore.
- Alignment: Processed reads were aligned to the human reference genome (hg19) using Bowtie2.
- Peak Calling: Accessible chromatin regions (ATAC-seq peaks) were identified using MACS2, with specific parameters to call broad peaks and identify peak summits. ENCODE blacklisted regions were excluded to avoid artifacts.
- Quality Control (QC): A four-metric QC process was implemented, including:
- Mean insert size and standard deviation (using Picard).
- TSS enrichment score.
- Fraction of Reads in Peaks (FRiP).
- Diagnostic plots were generated for visual assessment.
- Annotation: A wide range of genetic and epigenetic annotations were integrated, including:
- Super-enhancers and typical enhancers (from SEdb).
- Transcription Factor (TF) binding sites (predicted via TF footprint analysis and motif scanning using FIMO).
- Common Single Nucleotide Polymorphisms (SNPs) (from dbSNP).
- Risk SNPs (from GWAS Catalog and GWASdbv2.0).
- eQTLs (from PancanQTL, HaploReg, and GTEx).
- LD SNPs (calculated using PLINK).
- DNA methylation sites (from ENCODE).
- 3D chromatin interactions (from various sources like 4DGenome, Oncobase, NCBI, and 3D Genome Browser).
- Topologically Associating Domains (TADs) (from 3D Genome Browser).
- TF Footprint Analysis: This novel approach identifies TF binding sites by detecting patterns where TF binding prevents Tn5 transposase cleavage. It uses HINT for prediction and JASPAR for motif matching, providing TF protection scores and binding site counts.
- Associated Gene Identification: Genes linked to accessible chromatin regions were identified using closest, overlap, and proximal strategies via the ROSE tool.
- Visualization: Tools like ChIPseeker were used for peak annotation visualization, including pie charts of genomic feature distribution and peak heatmaps around TSS regions. A personalized genome browser (JBrowse) was also developed.
- Analysis Tools: ATACdb includes tools for "Differential-Overlapping-Region analysis" and "Overlapping accessible chromatin regions bound by two TFs."
Datasets
- Human ATAC-seq samples: Over 2200 samples were collected from NCBI GEO/SRA, with 1493 samples ultimately curated and processed, resulting in 52,078,883 accessible chromatin regions.
- Annotation Datasets:
- SEdb for super-enhancers and typical enhancers.
- JASPAR database for TF motifs.
- dbSNP for common SNPs.
- GWAS Catalog and GWASdbv2.0 for risk SNPs.
- PancanQTL, HaploReg, and GTEx for eQTLs.
- 1000 Genomes Project for phased genotype data.
- ENCODE for methylation sites and chromatin interaction data.
- 4DGenome and 3D Genome Browser for chromatin interactions and TADs.
Key Insights and Ideas
- The Power of Aggregation: By consolidating a massive number of ATAC-seq samples (over 1400, ~30x more than ENCODE), ATACdb provides unprecedented statistical power for analyzing human chromatin accessibility across diverse tissues and cell types.
- Beyond Peak Calling: TF Footprints: The inclusion of TF footprint analysis is a significant advancement, offering a more direct inference of TF binding and occupancy compared to simple motif scanning, which is crucial for understanding regulatory dynamics.
- Integrated (Epi)Genetic Context: The comprehensive integration of genetic (SNPs, eQTLs) and epigenetic (enhancers, methylation, chromatin interactions, TADs) annotations within accessible regions allows for a holistic view of regulatory mechanisms and their potential impact on disease.
- Standardized QC is Crucial: The emphasis on standardized quality control metrics (insert size, TSS enrichment, FRiP) is essential for ensuring the reliability and comparability of ATAC-seq data, a common challenge in the field.
- User-Centric Design: The database is designed with multiple query paths (genomic region, tissue, TF, gene), browse functionality, and integrated analysis tools, making it highly accessible and useful for researchers with varying needs.
Key Results
- ATACdb documents 52,078,883 accessible chromatin regions from 1493 human ATAC-seq samples.
- The database integrates a wide array of genetic and epigenetic annotations, providing rich contextual information for each accessible region.
- It offers accurate inference of TF footprints, a key feature for understanding TF binding and regulatory activity.
- The platform includes user-friendly tools for data querying, browsing, visualization, and analysis, such as differential region analysis and TF-bound region overlap analysis.
- A comparison with other databases (Table 1) highlights ATACdb's advantages in terms of sample size, QC guidelines, TF footprint analysis, and the breadth of integrated annotations.
Limitations and Future Work
- Species Extension: The current version focuses on human ATAC-seq data. Future work aims to extend the range of species covered and increase annotation information.
- Further Analysis Functions: The authors plan to add more practical analysis functions to the platform.
- Data Curation: While manual curation was performed, the sheer volume of data implies that ongoing efforts may be needed to refine data quality and consistency.
- Dynamic Nature of Chromatin: While TADs and chromatin interactions are included, capturing the dynamic nature of chromatin organization in real-time remains a challenge.
Summary updated