Overview

The traditional view of a microRNA (miRNA) locus producing a single, static "reference" mature sequence has been challenged by growing evidence of isomiRs — dynamic sequence variants that arise from each miRNA precursor arm. In plants, isomiRs result from imprecise cleavage during pre-miRNA processing and from post-transcriptional modifications, and act as important regulators alongside their canonical counterparts.

Despite extensive study of Arabidopsis miRNAs, systematic identification and annotation of isomiRs across tissues and conditions has remained limited. athisomiRDB was built to close that gap: it houses 20,764 isomiRs identified from 2,757 publicly available Arabidopsis small RNA-sequencing libraries, spanning more than 2,700 diverse samples, and allows exploration at the sample, miRNA, or isomiR level.

Each isomiR entry is annotated with its transcriptional origin, exclusivity (whether it maps uniquely within miRNA space or is shared with other genomic loci), known single-nucleotide variants from the 1001 Genomes Project, 3′ nontemplated post-transcriptional additions, predicted targets, and associations with plant traits — going well beyond the "one-locus-one-miRNA" paradigm.

How the Database Was Built

Arabidopsis sRNA-seq datasets were collected from the NCBI Sequence Read Archive (up to July 2022) using relevant keyword searches, yielding 2,757 datasets in total.

SRA Toolkit fastp v0.19.8 isoMiRmap miRBase R20 EDTA v1.8.4 1001 Genomes psRNATarget AtMAD BEDTools
  1. Raw data conversion — SRA files converted to FASTQ using the SRA Toolkit.
  2. Adapter trimming — 3′ adapters predicted and trimmed with fastp (v0.19.8).
  3. IsomiR identification — isoMiRmap profiled all wild-type isomiRs against a k-mer lookup table built from miRBase Release 20 precursors, each hairpin arm extended 6 nt beyond its annotated boundary to catch boundary-straddling isomiRs (18–26 nt inclusive).
  4. Exclusivity tagging — isomiRs classified as exclusive (unique to miRNA space) or ambiguous (also mapping elsewhere in the genome).
  5. Repeat annotation — transposable elements predicted with EDTA (v1.8.4) on the TAIR10 assembly, then intersected with every isomiR.
  6. Variant annotation — isomiRs cross-referenced against the 1001 Genomes Project (release 36) to flag known SNP-containing variants.
  7. Tissue specificity — a τ-value calculated per isomiR from expression data (higher τ = more tissue-specific).
  8. Target prediction — isomiR targets predicted with psRNATarget (2017 release, Schema v2), retaining the top 10 targets per isomiR.
  9. Multi-omics linkage — isomiR loci intersected with eQTL, GWAS, and other association data from AtMAD using BEDTools.

What You Can Explore

The database organizes 289 miRNA families across 6 major tissue types, 142 developmental stages, 68 experimental conditions, and 326 genotypes.

Browse

Systematic exploration by miRNA family, tissue, developmental stage, or genotype, with each isomiR's ID, sequence, genomic location, and exclusivity shown at a glance.

Search

Keyword/identifier, genomic location, source hairpin, BLAST sequence similarity, and variant-isomiR search engines.

Detail Pages

Each isomiR links to a page covering basic details, expression profile, variant information, and predicted targets/trait associations.

Notable findings behind the numbers

~97% of isomiRs overlapping a reference mature miRNA show variation at their 5′ and/or 3′ ends.
Uridylation (35.85%) and adenylation (27.7%) are the most common 3′ nontemplated additions.
201,789 isomiR–target interactions predicted across the full catalogue.
2,666 isomiR–QTL associations identified, including cis-/trans-eQTLs and GWAS-linked loci.

Case Studies from the Publication

IsomiR sorting into Argonaute (AGO) complexes

Using 28 public AGO-RIP datasets, 1,952 of the 20,764 catalogued isomiRs were found associated with an AGO complex. Distinct AGO complexes recruit largely non-overlapping isomiR subsets, and 5′ terminal nucleotide identity showed a pronounced bias per complex — pointing to selective, sequence-driven sorting.

miR775 isoforms and shared target pathways

Combining athisomiRDB target predictions with PARE-seq degradome data (via CleaveLand), isomiRs of miR775 were shown to cleave several targets beyond the previously reported canonical target GALT — suggesting isomiRs act as cooperative partners with the canonical miRNA in regulating related gene pathways.

Implementation

athisomiRDB runs on Apache HTTP Server, PHP, and MySQL on a CentOS 7 Linux server, with interactive visualizations built on Plotly and online sequence similarity search powered by BLAST. The site is tested on Chrome, Firefox, and Safari.

Apache 2.4.6 PHP MySQL 8.0.15 CentOS 7 Plotly BLAST v2.11.0

Citation

A T Vivek, Ajay Arya, Supriya P Swain, Shailesh Kumar, athisomiRDB: A comprehensive database of Arabidopsis isomiRs, Database, Volume 2024, 2024, baae115.

doi:10.1093/database/baae115

Team & Funding

Developed byA T Vivek and Ajay Arya (equal first authors), Supriya P Swain — Bioinformatics Laboratory, NIPGR
Corresponding authorDr. Shailesh Kumar — Bioinformatics Laboratory, National Institute of Plant Genome Research (NIPGR)
InstituteNational Institute of Plant Genome Research (NIPGR), Aruna Asaf Ali Marg, New Delhi 110067, India
FundingDepartment of Biotechnology (DBT), Government of India — project grants BT/PR40146, BT/PR40169, BT/PR40160, and BT/PR40261; core grant of NIPGR