TL;DR
Researchers have developed ENdb, a comprehensive, manually curated database of experimentally validated enhancers in humans and mice. This resource addresses the limitations of existing databases that primarily rely on high-throughput methods, offering a more reliable foundation for understanding gene regulation, disease mechanisms, and biological processes.
What Problem Does This Paper Address?
Understanding gene regulation is fundamental to biology, and enhancers play a crucial role by boosting gene transcription. These cis-regulatory elements are increasingly linked to human diseases and biological processes. While numerous enhancer databases exist, many focus on enhancers identified through high-throughput experimental techniques. These methods, while powerful for large-scale discovery, can suffer from higher rates of false positives. There's a clear need for a resource that meticulously curates enhancers supported by low-throughput, experimentally validated evidence, providing a more robust and reliable dataset for researchers. This is particularly important for dissecting complex regulatory mechanisms and their implications in health and disease.
Methods
The ENdb database was constructed through a rigorous manual curation process. Researchers reviewed approximately 1,590 published studies, meticulously extracting information on enhancers that met specific criteria: their genomic location had to be retrievable, and their enhancer activity must have been validated by at least one form of experimental evidence.
The data collection involved several key steps:
- Abstract Screening: Over 23,000 abstracts were initially screened using the keyword "enhancer."
- Publication Filtering: From the initial pool, 4,641 publications related to human and mouse enhancers were retained. These were further filtered to 1,590 based on the retrievability of genomic locations and experimental validation of enhancer activity.
- Detailed Extraction: The full text of these 1,590 publications was reviewed to extract detailed information, including:
- General Enhancer Information: PubMed ID, enhancer name, chromosome and positional information, experimental techniques used (e.g., luciferase reporter assay, EMSA, western blotting, 3C), experimental samples (cell line/tissue), enhancer class (low-throughput, low-+high-throughput), species, and reference genome (hg19/mm10).
- Enhancer-Related Information: Transcription factors (TFs) binding to enhancers, target genes, associated functions, and diseases.
- Evidence Strength Classification: Relationships between enhancers and target genes were categorized as "strong evidence" (supported by 3C, 4C, 5C, or CRISPR/Cas9) or "less strong evidence" (e.g., RNAi for TF knockdown).
- Standardization: Disease names were mapped to the Disease Ontology and MeSH databases, and target gene and TF names were standardized.
- Cross-referencing: Links to external resources like PubMed, HGNC, dbSNP, Disease Ontology, and MeSH were provided.
The database itself was developed using MySQL, with an interactive interface built with Bootstrap and JQuery. Graphical visualization was achieved using ECharts and D3, and JBrowse was integrated as the genome browser framework.
Datasets
The primary "dataset" for ENdb is the collection of 1,590 manually curated publications that provided experimentally validated enhancer information. The database itself contains:
- 737 experimentally validated enhancers (425 human, 312 mouse).
- 384 target genes.
- 263 transcription factors (TFs).
- 110 diseases.
- 153 functions.
- 34 SNPs.
The database also references external datasets and ontologies, including:
- PubMed database for initial screening.
- NCBI, HGNC, dbSNP for gene and SNP information.
- Disease Ontology and MeSH databases for disease classification.
- UCSC, GIVE, and Genome-Browse for genomic visualization.
Key Insights and Ideas
- The Value of Manual Curation: The paper strongly advocates for the importance of manual curation, especially for low-throughput experimental evidence, arguing it provides higher accuracy and reliability compared to high-throughput methods alone.
- Enhancer Location and Function: A significant portion of enhancers are located in distal intergenic regions and near promoters, suggesting complex regulatory interactions that can span considerable genomic distances. The finding that many enhancers near promoters regulate distant genes, rather than the closest one, highlights the non-linear nature of gene regulation.
- Enhancers in Disease: The database reveals a substantial number of enhancers associated with various diseases, underscoring their critical role in human pathology and offering potential targets for therapeutic intervention.
- Comprehensive Regulatory Network: By integrating enhancers with their target genes, TFs, and associated diseases, ENdb provides a holistic view of regulatory networks, facilitating a deeper understanding of gene expression control.
Key Results
- Database Size: The current release of ENdb contains 737 experimentally validated enhancers for human and mouse.
- Content Richness: The database links these enhancers to 384 target genes, 263 TFs, 110 diseases, and 153 functions.
- Uniqueness: The authors report that 629 (85.3%) of the enhancers in ENdb are uniquely collected within their database, indicating a significant contribution to the existing landscape of enhancer resources.
- Publication Trends: The number of publications involving enhancers has steadily increased over the past decade (2008-2018), highlighting the growing research interest in this field.
- Enhancer Location Distribution: Human enhancers are predominantly found in distal intergenic regions (39.7%), promoters (27.9%), and introns (22.5%). Mouse enhancers show a similar distribution, with distal intergenic regions (45.8%) being the most common.
- Promoter-Proximal Enhancers: Of enhancers located in promoter regions, only 40 (21%) were found to be within 2KB of the transcription start site, suggesting many regulate distant genes.
- Top Associations: The top 10 target genes, TFs, and diseases represent a significant portion of the curated data, indicating key players in enhancer-mediated regulation. For instance, MYC is a top target gene, and diseases like breast cancer and leukemia are highly represented.
Limitations and Future Work
- Manual Curation Bottleneck: The manual curation process, while ensuring high quality, is time-consuming and limits the sheer number of enhancers that can be included. The authors acknowledge that the number of enhancers in ENdb (737) is less than some other databases, but emphasize the superior quality of their data.
- Future Curation: The authors plan to continue manually curating newly validated enhancer associations and updating the database.
- Species Expansion: They also intend to expand the database to include experimentally supported enhancers from additional species.
- Data Integration: The paper suggests that ENdb has considerable potential to complement other enhancer-related databases, implying future work could involve further integration and comparative analysis.
Summary updated