Vienna RNA Predictor
RNA secondary structure prediction and target site accessibility scoring using the Vienna RNA package.
Overview
The Vienna RNA package is a well-established suite of algorithms for predicting RNA secondary structures based on minimum free energy (MFE) thermodynamic models. SPACER integrates Vienna RNA to assess target site accessibility — a key determinant of Cas13 guide efficacy.
Guides targeting regions of the transcript that are buried in stable secondary structures tend to show reduced cleavage activity. By scoring accessibility, SPACER penalizes guides aimed at highly structured regions and rewards those targeting single-stranded, accessible sites.
Minimum Free Energy (MFE) Calculation
For each candidate guide, SPACER extracts a local window of the target RNA centered on the binding site. The Vienna RNA fold algorithm computes the MFE structure of this window, producing both a free energy value (in kcal/mol) and a dot-bracket notation of the predicted structure.
Target window: AUGCUUACGAUCGAUCGAUCGUAGCUAGCUAG
MFE structure: ...((((...))))...(((....)))......
Free energy: -8.40 kcal/molAccessibility Scoring
SPACER converts the MFE structure into an accessibility score by calculating the fraction of unpaired nucleotides within the guide binding region. A fully single-stranded target site yields a score of 1.0, while a completely base-paired site yields 0.0.
| Accessibility | Interpretation | Effect on Score |
|---|---|---|
| 0.8–1.0 | Highly accessible, mostly single-stranded | Positive contribution |
| 0.5–0.8 | Partially accessible | Neutral to moderate contribution |
| 0.2–0.5 | Partially structured | Mild penalty |
| 0.0–0.2 | Highly structured, mostly base-paired | Significant penalty |
Window Parameters
The local folding window extends beyond the guide binding site to capture flanking structure context. The default window includes 50 nucleotides upstream and downstream of the target site. This prevents edge effects from artificially inflating accessibility scores.
Seed Accessibility
Beyond overall accessibility, SPACER computes seed accessibility — the average accessibility over the enzyme-specific seed region of the guide binding site. The seed region is critical for initial target recognition:
| Enzyme | Seed Region | Significance |
|---|---|---|
| Cas12 | Positions 0–5 (PAM-proximal) | Initial DNA unwinding and R-loop formation |
| Cas13 | Positions 5–9 (HEPN switch) | Decouples binding from cleavage activation |
A seed accessibility above 0.5 combined with an MFE above −3.0 kcal/mol indicates a favorable target site. The seed accessibility value feeds into the spacer_structure assay component of the composite score (default weight: 0.10).
Centroid Structure Prediction
In addition to the MFE structure, SPACER can compute the centroid structure via the partition function. The centroid minimizes the mean base pair distance to the full thermodynamic ensemble — making it more representative than MFE for complex structures like full crRNAs (spacer + direct repeat assemblies).
Both MFE and centroid predictions share the same base pair probability matrix and accessibility profiles. The centroid structure is particularly useful when evaluating how the direct repeat stem-loop interacts with the spacer sequence.
Integration with Composite Scoring
When the Vienna RNA predictor is enabled, structure analysis feeds into the spacer_structure assay component of the composite score. This component uses seed accessibility and MFE to penalize guides that target highly structured regions. The default weight for this component is 0.10.
Limitations
- MFE predictions represent the thermodynamically optimal structure, not necessarily the biologically active conformation
- Protein-RNA interactions and cellular context are not captured by thermodynamic models
- Pseudoknots are not predicted by the default Vienna RNA partition function
- Long-range interactions beyond the folding window may influence local structure