miRsponge: an R/Bioconductor package for the identification and analysis of miRNA sponge interaction networks and modules

Junpeng Zhang, Lin Liu, Taosheng Xu, Yong Xie, Chunwen Zhao, Jiuyong Li, Thuc Duy Le

TL;DR

This paper introduces miRspongeR, an R/Bioconductor package designed to streamline the identification and analysis of miRNA sponge interaction networks. It offers a comprehensive suite of tools for researchers to explore these complex regulatory mechanisms, accelerating discovery in fields like cancer research.

What Problem Does This Paper Address?

MicroRNA (miRNA) sponges are RNA molecules that can bind to and sequester miRNAs, thereby influencing gene expression. While their importance is increasingly recognized, our understanding of their complex functions remains limited. A significant bottleneck in this research area is the lack of user-friendly, integrated tools that allow researchers to efficiently identify miRNA sponges, analyze their interaction networks, and explore their biological implications. Existing computational methods are often scattered, making it difficult to compare them, apply them to new datasets, or select the most appropriate approach for experimental design. This paper addresses this gap by providing a unified R package that consolidates multiple identification methods and downstream analysis tools.

Methods

The core of miRspongeR is its pipeline, which consists of three main components:

  • Identification of miRNA sponge interactions: The package implements eight methods for this crucial step:
  • Seven stand-alone methods: These include miRHomology (based on shared miRNA homology), pc (positive correlation of expression), sppc (sensitivity partial Pearson correlation), hermes (mutual information and conditional mutual information), ppc (partial Pearson correlation), muTaME (considering MRE fraction, density, distribution, and proportion), and cernia (an updated version of muTaME with additional scores like hybridization energy and DT-Hybrid recommendation).
  • One integrative method (integrateMethod): This ensemble approach combines predictions from the seven stand-alone methods using a majority voting system to identify high-confidence interactions.
  • Identification of miRNA sponge modules: After identifying interactions, miRspongeR can group these interactions into modules (clusters of miRNA sponges) using four established algorithms: FN (fast hierarchical agglomeration), MCL (Markov cluster algorithm), LINKCOMM (link community detection), and MCODE (molecular complex detection).
  • Validation and analysis: This component allows for the assessment of predicted interactions and the exploration of their biological significance. It includes:
  • Validation of miRNA sponge interactions: Using curated databases of experimentally validated miRNA sponge interactions.
  • Functional enrichment analysis: Identifying enriched diseases, biological processes, and pathways associated with miRNA sponge modules using databases like Disease Ontology (DO), DisGeNET (DGN), Network of Cancer Genes (NCG), Gene Ontology (GO), KEGG, and Reactome.
  • Survival analysis: Assessing whether identified miRNA sponge modules can predict patient survival outcomes, using multivariate Cox models and Log-rank tests.

Datasets

The paper uses the BRCA dataset for a case study. This dataset comprises matched Level 3 IlluminaHiSeq miRNA and mRNA expression data from human breast invasive carcinoma, obtained from The Cancer Genome Atlas (TCGA). The data was restricted to 72 individuals with both tumor and normal tissues.

Additionally, the study utilizes:

  • miRTarBase v7.0: For experimentally validated miRNA-target interactions, specifically those supported by strong experimental evidence (Reporter assay or Western blot).
  • Ground truth data: This includes curated miRNA sponge interactions from sources like miRSponge, LncCeRBase, LncACTdb v2.0, and manually curated literature. This ground truth data is included within the miRspongeR package.

Key Insights and Ideas

  • The need for integration: The paper highlights that while numerous computational methods exist for identifying miRNA sponges, their fragmentation hinders progress. miRspongeR's strength lies in its integration of these methods into a single, user-friendly package.
  • Ensemble approach for confidence: The integrateMethod using majority voting is a key innovation, aiming to improve the reliability of predicted miRNA sponge interactions by requiring consensus among multiple methods.
  • Beyond interactions: Modules and function: Identifying miRNA sponge interactions is only the first step. The package's ability to identify modules and perform functional and survival analysis provides a more holistic view of their biological roles.
  • Data type matters: The paper implicitly emphasizes the importance of considering whether data is "dynamic" (changing across conditions) or "static" when choosing an identification method, with the integrateMethod handling both.

Key Results

  • The miRspongeR package successfully identifies miRNA sponge interactions, modules, and performs downstream analyses.
  • In the BRCA case study, the integrateMethod identified 87 high-confidence miRNA sponge interactions.
  • When validating against ground truth, only 1 miRNA sponge interaction (PTEN:ZEB1) was validated across all methods. However, the sppc method showed the best percentage of experimentally validated interactions.
  • Different module identification methods (FN, MCL, LINKCOMM, MCODE) yielded varying numbers of modules, with LINKCOMM identifying the largest number of modules, including overlapping ones.
  • Enrichment analysis revealed significant terms (DO, DGN, NCG, GO, KEGG, Reactome) associated with identified miRNA sponge modules, suggesting their involvement in various biological processes and diseases.
  • Survival analysis identified significant miRNA sponge modules capable of distinguishing high and low-risk BRCA samples.

Limitations and Future Work

  • Limited ground truth: The paper acknowledges that the ground truth for miRNA sponge interactions is still limited, making validation challenging. This necessitates further experimental validation and the continuous updating of curated databases.
  • Completeness of validated interactions: Experimentally validated miRNA-target interactions are far from complete. The authors suggest integrating multiple experimentally validated datasets (e.g., miRTarBase, TarBase, miRWalk) to obtain a more comprehensive list.
  • Future development: The authors plan to continuously include new methods in the package as research into miRNA sponges evolves, ensuring miRspongeR remains a state-of-the-art tool.

Summary updated

Abstract

Abstract Background A microRNA (miRNA) sponge is an RNA molecule with multiple tandem miRNA response elements that can sequester miRNAs from their target mRNAs. Despite growing appreciation of the importance of miRNA sponges, our knowledge of their complex functions remains limited. Moreover, there is still a lack of miRNA sponge research tools that help researchers to quickly compare their proposed methods with other methods, apply existing methods to new datasets, or select appropriate methods for assisting in subsequent experimental design. Results To fill the gap, we present an R/Bioconductor package, miRsponge , for simplifying the procedure of identifying and analyzing miRNA sponge interaction networks and modules. It provides seven popular methods and an integrative method to identify miRNA sponge interactions. Moreover, it supports the validation of miRNA sponge interactions and the identification of miRNA sponge modules, as well as functional enrichment and survival analysis of miRNA sponge modules. Conclusions This package enables researchers to quickly evaluate their new methods, apply existing methods to new datasets, and consequently speed up miRNA sponge research.
BibTeX
@article{zhang2019mirsponge,
  author = {Junpeng Zhang and Lin Liu and Taosheng Xu and Yong Xie and Chunwen Zhao and Jiuyong Li and Thuc Duy Le},
  title = {{miRsponge: an R/Bioconductor package for the identification and analysis of miRNA sponge interaction networks and modules}},
  journal = {BMC Bioinformatics},
  year = {2019},
  doi = {10.1186/s12859-019-2861-y},
}
Figure