What is RNAplex?
RNAplex is a fast algorithm for predicting RNA-RNA interactions, developed as part of the ViennaRNA Package. Unlike tools that compute full thermodynamic models of both inter- and intramolecular structures, RNAplex uses a simplified energy model that considers only intermolecular base pairs. This makes it 10-27 times faster than comparable tools like RNAhybrid while maintaining competitive accuracy.
The tool excels at scanning long target RNAs for potential binding sites of shorter query sequences. This asymmetric search pattern makes RNAplex particularly useful for microRNA target prediction, bacterial small RNA (sRNA) target identification, and antisense oligonucleotide design.
When combined with precomputed accessibility profiles from RNAplfold, RNAplex achieves prediction accuracy comparable to the more computationally intensive RNAup while running three orders of magnitude faster.
How does RNAplex work?
RNAplex decomposes predicted RNA duplexes into three structural elements: stacking pairs, bulge loops, and interior loops. The algorithm uses dynamic programming with four tables representing substructures ending in base pairs, interior loops, or bulges on either sequence.
Energy model
The simplified energy model ignores intramolecular structure within each RNA, focusing solely on the hybrid duplex. This approximation slightly overestimates loop energies but dramatically reduces computational complexity from O(N⁴) to O(N²), where N is sequence length.
Bulge loop penalties match the Turner thermodynamic model exactly for bulges up to 6 nucleotides. Larger loops use linear extrapolation.
Accessibility integration
Structured RNAs do not freely expose all nucleotides for interaction. RNAplex incorporates this by reading precomputed accessibility profiles that estimate the energetic cost of "opening" local secondary structure to permit intermolecular pairing.
The accessibility-enhanced mode:
- Precomputes local folding probabilities using RNAplfold
- Estimates the free energy cost to make each region accessible
- Adds opening costs to the interaction energy
Benchmarks show that accessibility-aware predictions (RNAplex, RNAup, IntaRNA) consistently outperform methods that ignore accessibility in both sensitivity and precision.
How to use RNAplex online
ProteinIQ provides cloud-based RNAplex predictions through a browser interface, eliminating the need to install the ViennaRNA Package locally.
Inputs
| Input | Description |
|---|---|
RNA Sequence 1 | First RNA sequence (query). FASTA format or raw nucleotides. |
RNA Sequence 2 | Second RNA sequence (target). FASTA format or raw nucleotides. |
Settings
| Setting | Description |
|---|---|
Temperature | Folding temperature (0-100, default 37). Affects thermodynamic calculations. |
Output columns
| Column | Description |
|---|---|
Interaction | The pair of sequence identifiers involved in the interaction. |
Hit # | Ranking of the interaction (1 = best). |
Structure | Dot-bracket notation showing the hybrid structure, with & separating the two strands. Paired bases shown as parentheses, unpaired as dots. |
Energy | Predicted binding free energy in kcal/mol. More negative values indicate stronger predicted binding. |
Interpreting results
The binding energy indicates thermodynamic favorability:
| Energy (kcal/mol) | Interpretation |
|---|---|
| < -20 | Very strong interaction |
| -10 to -20 | Strong interaction |
| -5 to -10 | Moderate interaction |
| > -5 | Weak interaction, may not be biologically relevant |
These thresholds are approximate. Biological relevance depends on cellular context, RNA concentrations, and competing structures.
When to use RNAplex vs other tools
The ViennaRNA Package includes several tools for RNA-RNA interaction prediction, each suited to different use cases:
- RNAplex: Fast screening of many potential interactions. Best for genome-wide target searches.
- RNAduplex: Simple duplex calculation without accessibility. Fastest option when local structure is not a concern.
- RNAup: Full accessibility calculation. More accurate but slower. Use for detailed analysis of specific candidates.
- RNAcofold: Computes full dimer structure including intramolecular pairs. Best for studying complexes where both RNAs fold together.
For typical workflows, RNAplex serves as an initial filter to identify candidates, which can then be analyzed more thoroughly with RNAup.
Applications
miRNA target prediction
MicroRNAs bind target mRNAs through partial complementarity, typically requiring a "seed" match at positions 2-8 of the miRNA. RNAplex identifies thermodynamically favorable binding sites that can be filtered for seed matches.
Bacterial sRNA regulation
Small regulatory RNAs in bacteria control gene expression by base-pairing with mRNA targets. These interactions often occur in 5' UTRs near ribosome binding sites. RNAplex enables rapid screening of potential sRNA-mRNA pairs.
Antisense oligonucleotide design
Designing antisense therapeutics requires identifying accessible regions in target RNAs. RNAplex predictions help select oligonucleotide sequences most likely to hybridize efficiently.
siRNA off-target prediction
Short interfering RNAs can silence unintended transcripts through partial complementarity. RNAplex screens for potential off-target binding sites across the transcriptome.
Limitations
RNAplex predictions assume thermodynamic equilibrium and do not account for:
- RNA-binding proteins that may block or facilitate interactions
- Cellular compartmentalization
- Kinetic effects and RNA folding pathways
- Post-transcriptional modifications
The simplified energy model trades some accuracy for speed. For high-confidence predictions of specific interactions, verify candidates using RNAup or experimental validation.
Related tools
- RNAduplex: Simple RNA hybridization without accessibility
- RNAup: Accessibility-aware interaction prediction (slower, more accurate)
- RNAcofold: Full dimer secondary structure prediction
- RNAfold: Single-sequence secondary structure prediction
- RNAplfold: Local folding probabilities and accessibility profiles
- ViennaRNA: Access all 14 ViennaRNA methods through a unified interface