EuRBPDB: a comprehensive resource for annotation, functional and oncological investigation of eukaryotic RNA binding proteins (RBPs)

Jian-You Liao, Bing Yang, Yu-Chan Zhang, Xiao-Juan Wang, Yushan Ye, Jing-Wen Peng, Zhi-Zhi Yang, Jie-Hua He, Yin Zhang, Kaishun Hu, De-Chen Lin, Dong YinNucleic Acids Res 2020 · DOI

TL;DR

Researchers have created EuRBPDB, the most comprehensive database to date for eukaryotic RNA binding proteins (RBPs). This resource integrates computational predictions and experimental data to provide detailed annotations, functional insights, and crucial cancer-related information for over 315,000 RBPs across 162 species, offering a powerful platform for RNA biology and cancer research.

What Problem Does This Paper Address?

RNA binding proteins (RBPs) are fundamental to gene regulation, influencing nearly every aspect of cellular life by interacting with RNA molecules. However, despite their critical roles in biological processes and their involvement in various diseases, including cancer, a complete and systematically annotated catalog of eukaryotic RBPs has been lacking. Existing databases were often limited in scope, species coverage, or the types of RBPs they included, making it difficult for researchers to get a comprehensive view of the RBP landscape. This paper addresses the need for a unified, extensive resource that not only identifies RBPs but also provides detailed functional and clinical context, particularly for cancer biology.

Methods

The researchers employed a multi-pronged approach to build EuRBPDB:

  • Computational RBP Identification: They utilized Hidden Markov Models (HMMER) to search for known RNA binding domains (RBDs) across the protein sequences of 162 eukaryotic species, sourced from Ensembl. This method identified "canonical" RBPs.
  • Integration of RBPome Data: To capture "non-canonical" RBPs (those lacking known RBDs but experimentally shown to bind RNA), they integrated data from large-scale RBPome studies (e.g., RNA interactome capture, OOPS, CLIP-Seq) across six key species (human, mouse, zebrafish, fly, worm, and yeast).
  • Data Curation and Annotation: Information for each identified RBP was gathered from various public databases, including NCBI, Ensembl, STRING (for protein-protein interactions), KEGG (for pathways), and GeneCards.
  • Cancer-Specific Analysis: RBPs were systematically investigated in the context of cancer biology by mining published literature, analyzing protein-protein interaction networks, and integrating large-scale omics data (RNA-Seq, copy number variation, mutations, and survival data) from TCGA.
  • Web Interface Development: A user-friendly web interface was designed to allow for easy browsing, searching, and interactive exploration of the database, including a dedicated cancer web interface.

Datasets

  • Protein Sequences: From Ensembl database (release 96) for 162 eukaryotes.
  • RNA Binding Domains (RBDs): Curated list based on Pfam and previous studies.
  • RBPome Datasets: Large-scale experimental data from 21 published works for human, mouse, zebrafish, yeast, fly, and worm.
  • Public Databases: NCBI, Ensembl, STRING, KEGG, GeneCards, ENCODE, GEO, Sequence Read Archive (SRA), GTEx, and TCGA.
  • Cancer Data: TCGA data for differential expression, somatic mutations, copy number variations (CNVs), and survival analysis.
  • L1000 Assay Data: From GEO (GSE92742 and GSE70138) for drug effects on RNA profiles.
  • eCLIP Datasets: From ENCODE for RNA binding sites.

Key Insights and Ideas

  • The RBPome is Vast and Diverse: The study reveals a significantly larger repertoire of RBPs than previously appreciated, highlighting the importance of integrating both computational predictions and experimental RBPome data for a comprehensive view.
  • Non-Canonical RBPs are Crucial: The identification of a substantial number of non-canonical RBPs underscores that RNA binding is not solely dependent on canonical domains, and other protein structures or regions play significant roles.
  • RBPs are Deeply Implicated in Cancer: The systematic analysis reveals that a large proportion of RBPs exhibit cancer-associated alterations (expression changes, mutations, CNVs, survival correlations), suggesting their critical involvement in tumorigenesis and progression. This provides a valuable resource for identifying novel cancer biomarkers and therapeutic targets.
  • A Unified Resource is Essential: The creation of EuRBPDB addresses a critical need for a centralized, comprehensive, and user-friendly platform that consolidates diverse data types for RBPs, facilitating research across multiple disciplines.

Key Results

  • Identification of 315,222 RBPs: This includes 311,571 canonical RBPs (grouped into 6,368 ortholog groups) and 3,651 non-canonical RBPs across 162 eukaryotic species.
  • Largest Eukaryotic RBP Database: EuRBPDB is presented as the most comprehensive eukaryotic RBP database currently available.
  • Extensive Annotation: RBPs are annotated with basic information, functional data (PPI, GO, pathways), and cancer-relevant information.
  • Cancer Alterations Identified:
    • 1,361 RBPs show aberrant expression in at least one cancer type.
  • 2,900 RBPs harbor nonsense/missense mutations (1,761 in RBD regions).
  • 2,851 RBPs have genomic deletions or amplifications.
  • 2,897 RBPs exhibit significant survival correlation.
  • Validation: EuRBPDB successfully integrated the vast majority of RBPs found in other existing databases, demonstrating its comprehensiveness and accuracy.

Limitations and Future Work

  • Limitations of RBPome Technologies: The authors acknowledge that RBPome technologies, while powerful, are not exhaustive and can be limited by context-dependent RNA binding, restricted expression patterns, technical limitations, and low sensitivity.
  • Ongoing Updates: The researchers plan to continuously update and maintain the RBP repertoire and annotations as new data becomes available.
  • Integration of Additional Omics Data: Future work includes integrating more omics datasets (e.g., CLIP-seq, RNA-Seq) from public repositories to further enhance the understanding of RBP function and regulation.
  • RBPredictor Web-Server: A web-based tool, RBPredictor, has been developed to help users determine if a protein of interest from any eukaryote is a putative RBP, based on the RBD sets used in this study.

Summary updated

BibTeX
@article{liao2020eurbpdb,
  author = {Jian-You Liao and Bing Yang and Yu-Chan Zhang and Xiao-Juan Wang and Yushan Ye and Jing-Wen Peng and Zhi-Zhi Yang and Jie-Hua He and Yin Zhang and Kaishun Hu and De-Chen Lin and Dong Yin},
  title = {{EuRBPDB: a comprehensive resource for annotation, functional and oncological investigation of eukaryotic RNA binding proteins (RBPs)}},
  journal = {Nucleic Acids Res.},
  year = {2020},
  doi = {10.1093/nar/gkz823},
}
Figure