What is RNAplfold?
RNAplfold computes local base pair probabilities for RNA sequences using a sliding window approach. Part of the ViennaRNA Package, it calculates the mean probability that any two nucleotides within a defined span form a base pair, averaged across all sequence windows containing that pair.
The algorithm scales efficiently with sequence length, using O(n + L²) memory and O(n · L²) CPU time, where n is sequence length and L is window size. This makes RNAplfold practical for scanning entire genomes to identify regions with stable local secondary structures.
Beyond pairing probabilities, RNAplfold can compute accessibility profiles: the probability that stretches of consecutive nucleotides remain unpaired. These profiles are essential for predicting where small RNAs, antisense oligonucleotides, or RNA-binding proteins might interact with a target sequence.
How does RNAplfold work?
RNAplfold extends the McCaskill partition function algorithm to local structure prediction. For each position in the sequence, it computes the probability that a base pair (i, j) forms, considering only structures where both positions fall within the same window of size L.
The key insight is averaging: rather than reporting probabilities for a single window, RNAplfold computes the mean pairing probability across all windows containing positions i and j. This averaging smooths out boundary effects and provides more robust estimates of local structural propensity.
The span parameter restricts the maximum distance between paired bases. Setting span smaller than window size focuses the analysis on short-range interactions, which dominate most biologically relevant local structures.
Applications
Local base pair probabilities from RNAplfold support several research applications:
- MicroRNA target prediction: Accessible regions in mRNA 3' UTRs indicate potential miRNA binding sites
- siRNA design: Identifying unpaired stretches in target mRNAs where siRNAs can bind effectively
- Riboswitch analysis: Detecting structured aptamer domains in long regulatory regions
- Genome scanning: Screening for conserved structured elements like miRNA precursors
How to use RNAplfold online
ProteinIQ hosts RNAplfold on cloud infrastructure, enabling analysis of long RNA sequences directly in the browser without local ViennaRNA installation.
Input
| Input | Description |
|---|---|
RNA Sequences | One or more RNA sequences in FASTA format or as plain text. Supports sequences up to 50 MB. |
Settings
Probability options
| Setting | Description |
|---|---|
Window size | Size of the averaging window (20-200, default 70). Larger windows capture longer-range interactions but increase computation time. |
Maximum span | Maximum distance between paired bases (10-100, default 40). Must not exceed window size. |
Temperature | Folding temperature in degrees Celsius (0-100, default 37). Affects Boltzmann weighting of structures. |
Output
Results are returned as a spreadsheet with one row per input sequence:
| Column | Description |
|---|---|
Sequence ID | Identifier from FASTA header or auto-generated |
Length | Sequence length in nucleotides |
Window Size | Window size used for computation |
Span | Maximum span setting |
Computed | Whether computation completed successfully |
Detailed probability profiles are available in the downloadable output files, including position-specific pairing and accessibility data.
Choosing window and span parameters
The relationship between window size and span determines what structures RNAplfold can detect:
- Short hairpins (miRNA precursors, tRNAs): Window 70-100, span 30-50
- Longer structures (riboswitches, regulatory elements): Window 150-200, span 60-100
- Genome scanning (quick overview): Window 70, span 40 (defaults)
Larger windows and spans increase sensitivity to longer-range pairings but significantly increase runtime. For initial screening, the default parameters work well; refine them based on the specific structural elements being sought.
Limitations
RNAplfold assumes single-stranded RNA folding in isolation. It does not account for:
- RNA-protein interactions that constrain structure
- Pseudoknots (like all ViennaRNA tools using the standard recursions)
- Co-transcriptional folding kinetics
- Intermolecular base pairing
For RNA-RNA interaction prediction, use RNAplex or RNAup instead.
Related tools
- RNAfold: Global minimum free energy structure prediction
- RNALfold: Local structure prediction with MFE (structures, not probabilities)
- RNAplex: Fast RNA-RNA interaction prediction using accessibility profiles
- RNAup: RNA interaction prediction with explicit accessibility costs
- RNAsubopt: Enumerate suboptimal structures within an energy range
- ViennaRNA: Access all 14 ViennaRNA methods through a unified interface