TL;DR
Researchers have developed CATA, a comprehensive database that consolidates and annotates chromatin accessibility data across 23 cancer types. This resource aims to empower cancer researchers by providing a unified platform for exploring regulatory elements, identifying potential driver genes, and analyzing their clinical relevance, ultimately accelerating cancer research.
What Problem Does This Paper Address?
Understanding how DNA accessibility changes within a cell's genome is crucial for deciphering gene regulation, especially in the context of cancer. Accessible chromatin regions (CARs) are dynamic and often harbor regulatory elements that control gene expression. While methods like ATAC-seq have revolutionized our ability to map these regions, existing databases often lack comprehensive, cancer-specific annotations or integrated analytical tools. This fragmentation makes it challenging for researchers to efficiently explore the complex regulatory landscape of cancer, identify key regulatory elements, and link them to clinical outcomes. CATA aims to bridge this gap by providing a centralized, well-annotated, and user-friendly resource for cancer chromatin accessibility data.
Methods
The core of CATA is built upon publicly available ATAC-seq data from The Cancer Genome Atlas (TCGA), covering 410 tumor samples across 23 cancer types. The raw ATAC-seq data underwent a standardized processing pipeline, including alignment to the hg38 human reference genome, duplicate removal, and peak calling using MACS2. The resulting CARs were then extensively annotated using a variety of computational tools and integrated datasets.
Key annotation strategies include:
- Genomic and Epigenetic Annotations: Using BEDTools, CARs were annotated with common single nucleotide polymorphisms (SNPs), risk SNPs, copy number variations (CNVs), somatic mutations, transcription factor binding sites (TFBS), methylation data, enhancers (from FANTOM5), and CRISPR/Cas9 target sites.
- Gene Annotation: The ROSE genemapper method was employed to identify genes associated with CARs based on their proximity (overlap, within 50kb, or closest gene).
- TF Binding and Motif Analysis: TF binding regions were identified using FIMO predictions, and motif changes were analyzed by comparing binding affinities to mutated motifs using the R package
atSNP. - eQTL Analysis: Human eQTL data from GTEx, HaploReg, and PancanQTL were integrated to map eQTL-related SNPs to CARs and identify SNP-regulated genes.
- Pathway Analysis: CATA integrates 10 pathway databases (e.g., KEGG, Reactome) to allow users to perform enrichment analysis on genes associated with CARs.
The database itself is implemented using MySQL and Java with a Springboot framework, featuring a Bootstrap-based front end and Echarts for network visualization.
Datasets
- TCGA ATAC-seq data: Used as the primary source for chromatin accessibility regions across 23 cancer types and 410 samples.
- dbSNP (version 1.50): For common SNPs.
- 1000 Genomes Project Phase 3: For linkage disequilibrium (LD) SNPs.
- GWAS Catalog and GWASdb v2.0: For risk SNPs.
- TRANSFAC and JASPAR: For motif analysis.
- FANTOM5: For enhancer data.
- GTEx v5.0, HaploReg, and PancanQTL: For eQTL data.
- UCSC XENA: For methylation, RNA expression, somatic mutation, copy number variation, and clinical data.
No new benchmark datasets were introduced; instead, CATA integrates and annotates existing data.
Key Insights and Ideas
- Cancer-Specific Regulatory Landscapes: The database highlights that while some CARs are shared across cancer types (e.g., PDL1 promoter), many are highly specific to particular cancers, underscoring the need for cancer-type-specific analysis.
- Integrated Annotation for Deeper Insights: By combining diverse annotations (SNPs, eQTLs, TFBS, enhancers, etc.) with chromatin accessibility data, CATA enables researchers to explore the functional implications of CARs in a more holistic manner.
- Bridging Accessibility to Clinical Outcomes: The inclusion of clinical data and survival analysis tools allows researchers to directly investigate the prognostic value of genes associated with CARs, potentially identifying novel driver genes.
- User-Friendly Exploration: The development of intuitive search interfaces, interactive tables, and visualization tools (like the GIVE genome browser) makes complex genomic data more accessible to a broader research community.
Key Results
- CATA stores 2,991,163 CARs from 410 tumor samples across 23 cancer types.
- The database integrates annotations for 1,398 TFs within these CARs.
- Extensive annotation data includes: 38,063,729 common SNPs, 264,514 risk SNPs, 2,886,133 eQTLs, 9,408 CNVs, 8,631 SNVs, 65,424 enhancer locations, and 5,797,266 conserved TFBS.
- The platform provides user-friendly tools for data retrieval, visualization, and analysis, including pathway enrichment and survival analysis.
Limitations and Future Work
- Limited Data Types: The current version primarily focuses on ATAC-seq data. Future versions are planned to include ChIP-seq data, single-cell ATAC-seq data, and additional practical analysis tools.
- Data Updates: While the database is regularly maintained, the authors plan to incorporate new cancer open data and annotation data annually.
- Specific Cancer Types: While 23 cancer types are covered, the authors acknowledge that not all cancer types are represented.
Future work will focus on expanding the data types, enhancing analytical capabilities, and potentially incorporating more advanced machine learning approaches for predicting regulatory elements and their roles in cancer.
Summary updated