PMTED: a plant microRNA target expression database

Xiuli Sun, Boquan Dong, Lingjie Yin, Rongzhi Zhang, Wei Du, Dongfeng Liu, Nan Shi, Aili Li, Yanchun Liang, Long MaoBMC Bioinformatics 2013 · DOI

TL;DR

This paper introduces PMTED, a novel database that integrates plant microRNA (miRNA) target gene expression data from a vast collection of microarray experiments. By providing tools for querying, analyzing, and visualizing this data, PMTED aims to help researchers uncover the functional significance of miRNA-target interactions and generate new hypotheses for experimental validation.

What Problem Does This Paper Address?

MicroRNAs (miRNAs) are crucial regulators of plant development and stress responses, exerting their functions by interacting with target messenger RNAs (mRNAs). Identifying these miRNA targets is key to understanding their roles. While bioinformatics tools can predict potential targets, validating these predictions and understanding their contextual significance in biological processes has been challenging. This is largely due to the scattered nature of vast amounts of plant gene expression data, primarily from microarray experiments, which are often embedded in large public databases and require significant pre-processing and analytical tools to extract meaningful insights. Researchers need a centralized resource that not only lists predicted targets but also links them to their expression patterns under various biological conditions.

Methods

The authors developed PMTED by curating and integrating data from numerous plant microarray experiments. The core of their approach involves:

  • Data Acquisition and Pre-processing: Microarray data from Affymetrix platforms were retrieved from the Gene Expression Omnibus (GEO). Experiments lacking raw intensity data or biological replicates were excluded. Raw data were pre-processed using the Robin program, with normalization performed using RMA and MAS 5.0.
  • MiRNA Target Prediction: A set of 2,393 mature miRNA sequences from the Plant MicroRNA Database (PMRD) was used to predict target genes. This prediction utilized gene models or CDSs for sequenced genomes and updated cDNA sequences or EST contigs for others. A custom Perl script was employed, with a stringent cut-off threshold (score ≤ 4, E-value = 10) based on established criteria for validated miRNA-target pairings.
  • Data Curation and Classification: High-quality experiments were screened and classified into three major bioprocesses: abiotic stress, biotic stress, and development. These were further categorized into 55 unique terms representing specific experimental conditions.
  • Meta-analysis and Network Construction: Differentially expressed genes were identified using the limma package. For meta-analysis, experiments with similar biological processes were linked using common terms. The strength of these links was quantified by the coefficient of variation (CV) of expression values. The resulting meta-networks, representing relationships between miRNAs, bioprocesses, conditions, and species, were visualized using the Cytoscape Web program.

Datasets

  • Microarray Data: Data from 3,492 Affymetrix arrays across 311 experiments were used. These experiments covered 12 plant species, including Arabidopsis thaliana, Glycine max, Zea mays, Oryza sativa, and others. Experiments were classified into five categories: stimulus, development, mutation, small RNA and epigenetic, and others.
  • MiRNA Sequences: 2,393 mature miRNA sequences were downloaded from the Plant MicroRNA Database (PMRD).
  • Existing Databases: Data was sourced from Gene Expression Omnibus (GEO) and miRBase.

Key Insights and Ideas

  • Bridging the Gap: PMTED effectively bridges the gap between raw microarray data and actionable biological insights by providing a curated, searchable, and analyzable resource for miRNA targets.
  • Contextual Significance: The database emphasizes the "contextual significance" of miRNA targets by linking them to specific biological processes and experimental conditions, allowing researchers to infer functional roles.
  • Meta-analysis for Hypothesis Generation: The meta-network module is a powerful tool for generating novel hypotheses by revealing global relationships and potential connections between miRNAs, targets, and biological events across different species and experiments.
  • Facilitating Cross-Species Transfer: PMTED aims to facilitate the transfer of knowledge from well-studied model plants (like Arabidopsis) to less studied crop species by integrating data from a diverse range of plants.

Key Results

  • PMTED hosts information for 1,897 miRNAs and their 5,449 predicted target genes.
  • The database integrates data from 311 microarray experiments comprising 3,492 arrays across 12 plant species.
  • A total of 92 experiments were curated and classified into three major bioprocesses and 55 unique terms.
  • The target prediction pipeline, using a stringent cut-off, generated fewer false positives compared to the psRNATarget program, with a higher proportion of degradome-validated targets in rice.

Limitations and Future Work

  • Current Scope: The initial version of PMTED focuses on abiotic, biotic, and development experiments. The authors plan to expand its scope to include more species and experimental conditions in the future.
  • Data Compatibility: Pipelines are being developed to ensure data compatibility from other microarray platforms.
  • Integration of New Technologies: Future versions will incorporate data from next-generation sequencing technologies (e.g., RNA-seq), expression data of miRNAs themselves, and relevant literature.
  • User-Defined Predictions: While the database provides a portal for users to predict targets for novel miRNAs, the primary focus is on curated data.

Summary updated

BibTeX
@article{sun2013pmted,
  author = {Xiuli Sun and Boquan Dong and Lingjie Yin and Rongzhi Zhang and Wei Du and Dongfeng Liu and Nan Shi and Aili Li and Yanchun Liang and Long Mao},
  title = {{PMTED: a plant microRNA target expression database}},
  journal = {BMC Bioinformatics},
  year = {2013},
  doi = {10.1186/1471-2105-14-174},
}
Figure