PhyloRNA: a database of RNA secondary structures with associated phylogenies

PhyloRNA is a curated meta-database that bridges RNA secondary structures with comprehensive phylogenetic annotations and structural descriptors, enabling large-scale comparative and evolutionary analyses of RNA molecules.

The Gist

Imagine you are a detective trying to solve a massive mystery: How do tiny molecules called RNA evolve and change over time?

RNA is like the "messenger" inside our cells. It carries instructions (like a recipe) and also acts as a tool to build things. But here's the tricky part: RNA isn't just a straight line of letters (like a sentence). To do its job, it folds itself into complex 3D shapes, like origami.

Scientists know that while the letters in the RNA sentence might change wildly between different species (like a human vs. a bacteria), the folded shape often stays the same because that shape is what actually does the work.

The Problem: A Library Without a Catalog

For years, scientists have had libraries full of these RNA "origami" shapes. But there was a huge problem: The books were unorganized.

If you wanted to find all the RNA shapes belonging to "Bacteria" or "Mammals," you couldn't just ask the librarian. You'd have to manually check every single book, look up the organism's name in a separate encyclopedia, and hope you didn't make a mistake. Different encyclopedias (databases) even disagreed on what to call the same animal! It was like trying to sort a pile of mixed-up puzzle pieces from 1,000 different puzzles, where everyone calls the pieces by different names.

The Solution: PhyloRNA (The Smart Organizer)

Enter PhyloRNA. Think of PhyloRNA as a super-smart, automated librarian that has finally organized the entire RNA library.

Here is how it works, using some simple analogies:

1. The "Universal Translator" (Taxonomy)
Imagine you have a list of names written in five different languages (English, French, German, etc.). If you want to group all the "dogs" together, you have to translate every name first.
PhyloRNA does this automatically. It takes every RNA molecule and tags it with its family tree (its phylogeny) from five different major classification systems (like NCBI, SILVA, and GTDB).

• Why it's cool: You can ask, "Show me all the RNA from Bacteria," and it instantly gives you the list, even if you switch your question to use a different classification system. No more manual translation!

2. The "Shape Shifter" (Structural Abstractions)
Looking at a complex RNA fold is like looking at a tangled ball of yarn. It's hard to see the pattern.
PhyloRNA offers three ways to simplify the view, like looking at a map:

• The "Shape" View: It removes the loose ends (unpaired letters) and just shows the main loops. It's like looking at the silhouette of a building.
• The "Core" View: It strips away the decorations to show the main structural pillars.
• The "Core Plus" View: A slightly more detailed version of the pillars.
  These tools help scientists spot patterns. For example, they might realize that "All the RNA from the Bacteria kingdom looks like a specific type of knot, while Eukaryotes look like a different knot."

3. The "Magic Download" Button
In the old days, if a scientist wanted 1,000 RNA files to study, they had to download them one by one, rename them, and add the family tree info manually.
With PhyloRNA, you just click a few boxes (e.g., "I want tRNA," "I want it from Bacteria," "I want it in a specific format"), and the system spits out a neat folder with 1,000 files, all perfectly labeled and ready to go.

Real-World Superpowers

The paper shows three ways this tool helps scientists:

• Rebuilding Family Trees: It helps scientists quickly gather the right data to figure out how different species are related to each other based on their RNA shapes.
• Comparing Classifications: It lets scientists see how different "encyclopedias" classify the same organism. Maybe one says a creature is a "Bacteria" and another says "Proteobacteria." PhyloRNA lets you see both views side-by-side to understand the differences.
• Finding Hidden Patterns: By using the "Shape" simplifications, scientists found that certain RNA shapes are almost exclusively found in specific groups of life, helping them understand how evolution shapes these molecules.

The Bottom Line

PhyloRNA is a free, online database that acts as a bridge between RNA shapes and evolutionary history. It takes a messy, manual, error-prone job and turns it into a simple, automated search. It allows scientists to stop worrying about organizing data and start focusing on the big questions: How did life evolve, and how do these tiny molecular machines work?

You can visit their "library" at: https://bdslab.unicam.it/phylorna/

Technical Summary

1. Problem Statement

The central challenge addressed by this work is the lack of a unified resource that systematically links RNA secondary structures with curated phylogenetic classifications.

• Conservation vs. Sequence: While RNA secondary structure is known to be more evolutionarily conserved than primary sequence, making it vital for comparative and phylogenetic studies, existing databases fail to integrate structural data with rigorous taxonomy.
• Fragmentation: Existing resources (e.g., RNA STRAND, BpRNA, Rfam, CRW) either focus on sequence-centric data, provide only consensus structures, or lack explicit, curated taxonomic annotations.
• Inconsistency: Taxonomic resources (ENA, SILVA, LTP, NCBI, GTDB) differ in scope, naming conventions, and rank definitions. Manually mapping large sets of RNA molecules to multiple taxonomies is labor-intensive, error-prone, and hinders large-scale comparative analyses.
• Gap: No existing tool allows users to query RNA structures based on multiple curated phylogenies simultaneously or to download large datasets with consistent taxonomic metadata in various formats.

2. Methodology

PhyloRNA is implemented as a three-tier web application designed to automate the acquisition, processing, and integration of RNA structural and taxonomic data.

System Architecture

• Frontend: A Flask-based web interface served via Apache/WSGI.
• Backend: An application layer handling search, filtering, and download logic.
• Data Layer: A MongoDB database storing metadata and a GridFS mechanism for storing secondary structure files.
• Update Pipeline: Scheduled offline scripts periodically fetch data from external sources, process them, and update the database.

Data Acquisition & Processing Workflow

1. Sources: Data is aggregated from public repositories including the Protein Data Bank (PDB), CRW/CRW2, tmRNA, and SPRNA.
2. Structure Derivation:
   • Structures from PDB are derived using RNAView to extract base-pairing information from 3D coordinates.
   • Predicted structures are taken directly from source databases.
3. Standardization: All structures are converted into a unified internal representation and exported via the TARNAS tool into multiple formats:
   • (Extended) Dot-Bracket Notation
   • BPSEQ
   • CT (Connectivity Table)
   • FASTA (sequence only)
4. Descriptor Computation: Scripts calculate structural descriptors including sequence length, base pair counts, unpaired nucleotides, pseudoknot order, and genus.
5. Taxonomic Integration: Python scripts query five curated taxonomy systems (ENA, SILVA, LTP, NCBI, GTDB) for every RNA molecule. The resulting taxonomic hierarchy (from domain to species) is stored as nested JSON documents linked to the RNA entry.
6. Abstraction Levels: The system computes three levels of structural abstraction (Shape, Core, Core Plus) to simplify complex structures for comparative analysis.

3. Key Contributions

• PhyloRNA Database: The first curated meta-database that explicitly associates individual RNA secondary structures with five distinct, curated phylogenetic systems.
• Multi-Taxonomy Support: Users can switch between taxonomies (e.g., SILVA vs. GTDB) via the search interface without manual re-mapping, enabling direct comparison of how different classification systems affect evolutionary analyses.
• Rich Structural Descriptors: Beyond standard sequences, the database provides:
  • Pseudoknot metrics: Order and presence.
  • Topological measures: Genus.
  • Structural Abstractions: Shape, Core, and Core Plus (simplified representations that preserve biological features while reducing complexity).
• Validated Structures: Inclusion of over 4,000 secondary structures derived from experimentally resolved 3D models (PDB), filterable by "Is Validated = Yes."
• Flexible Export: Users can download search results in multiple formats (CSV, Dot-Bracket, BPSEQ, etc.) with or without headers, annotated with the specific taxonomy chosen for the query.

4. Results & Use Cases

The paper demonstrates the utility of PhyloRNA through three specific use cases:

1. Phylogenetic Reconstruction:

   • PhyloRNA automates the creation of benchmark datasets. A previous study requiring manual retrieval of phylum assignments for 5S, 16S, and 23S rRNAs could now be assembled automatically with access to multiple taxonomies, significantly reducing error and effort.

2. Comparative Taxonomy Analysis:

   • The authors retrieved 5S rRNAs and analyzed their clustering performance (using hierarchical clustering and ASPRAlign distance) across the five supported taxonomies.
   • Finding: Clustering performance (measured by Rand index, homogeneity, and completeness) varied significantly between taxonomies. For instance, GTDB and LTP showed higher agreement with known superkingdom labels than ENA or NCBI for specific datasets. This highlights the necessity of having access to multiple curated systems to avoid bias in evolutionary studies.

3. Structural-Phylogenetic Correlation:

   • Using Core Plus abstractions, the authors analyzed 264 5S rRNAs.
   • Finding: Two dominant Core Plus motifs accounted for ~86% of the dataset. One motif was predominantly associated with Eukaryota (81.1%), while the other was strongly associated with Bacteria (94.3%). This demonstrates the ability to perform large-scale statistical analyses linking specific structural motifs to taxonomic groups.

5. Significance

• Bridging the Gap: PhyloRNA fills a critical void by integrating structural biology with evolutionary taxonomy, a combination previously unavailable in a single, queryable resource.
• Scalability: It eliminates the manual bottleneck of mapping thousands of RNA molecules to taxonomies, enabling large-scale "structure-centric" evolutionary studies.
• Reproducibility & Flexibility: By providing data from multiple taxonomic sources and standardized structural formats, it allows researchers to test the robustness of their phylogenetic conclusions against different classification standards.
• Future Potential: The platform is designed for extensibility, with planned additions for structural pattern searching (motif discovery) and the inclusion of more RNA families and PDB-derived structures.

Availability: The database is publicly accessible at https://bdslab.unicam.it/phylorna/ and is regularly updated.