RNAExplorer – Help & Tutorial

RNAExplorer is an interactive web application for exploring, comparing, and analysing small RNA sequences using both deep learning embeddings and biologically interpretable features.

1. Visualise Sequences

RNAExplorer allows you to visualise small RNA sequences. Sequences can be provided using:

  • Plain text input
  • FASTA files
  • CSV files (sequence column must be named sequence)

After entering sequences, click “Overlay / Visualize” to project them onto the reference dataset. In production mode, the system automatically detects exact matches within the reference database when available. In local mode, the system generate the RNA-FM embeddings for user provided sequences with interpretable features and project then on top of reference sequences from database.

2. RNA-FM Embedding

The top plot displays the RNA-FM embedding, generated using a pretrained RNA-FM model. This view places your sequences within the landscape of known siRNA, miRNA, and piRNA reference sequences.

  • Each point represents a reference sequence
  • Colours indicate RNA class
  • Your uploaded sequences are highlighted
  • Query matches appear as red markers

This plot provides a global, similarity-based view of how your sequences relate to known small RNA classes.

3. Interpretable Feature Embedding

The bottom plot shows an interpretable feature embedding based on biologically meaningful sequence properties rather than deep learning representations.

Features include:

  • Sequence length
  • GC percentage
  • GC skew
  • AT/AU skew
  • Mononucleotide composition (A, C, G, U)
  • 2-mer and 3-mer frequencies

Selecting or deselecting feature checkboxes will only update the interpretable feature embedding. The reference dataset is re-projected using the selected features, allowing focused exploration of which properties drive separation between RNA classes.

4. Query Sequence Search

The Query sequence panel allows you to paste a single RNA sequence and search the reference database.

Clicking “Find in reference” triggers two analyses:

  • Exact matching sequences and reference sequences containing the query fragment
  • Most similar sequences based on RNA-FM embedding similarity

Matching sequences are highlighted on both embedding plots and reported in detailed tables below.

5. Interactive Tables

RNAExplorer provides integrated tables for detailed inspection and downstream analysis.

Each table includes:

  • RNA type (siRNA, miRNA, piRNA)
  • Sequence
  • Species (linked to original source)
  • Length, GC %, skew metrics
  • Mononucleotide composition

Users can filter by RNA type, show or hide columns, sort values, and download the visible data as CSV files.

6. Interactive Points and External Links

Individual sequences shown in the plots are fully interactive. Clicking a point redirects you to the original data source from which the sequence was derived, providing access to additional metadata and biological context.

7. Downloads

RNAExplorer provides multiple options for downloading data and figures generated during analysis.

Downloads include only the data currently displayed in the interface.