ProteinIQ

RNAdos

Optimal RNA sequence design

What is RNAdos?

RNAdos is an iterative RNA sequence design tool from the ViennaRNA Package. Given a target secondary structure in dot-bracket notation, it searches for RNA sequences that fold into that exact structure. Unlike single-pass inverse folding, RNAdos performs multiple optimization rounds to produce sequences with better folding properties and higher success rates.

The inverse folding problem is computationally challenging: there may be millions of sequences compatible with a given structure, but only some will reliably fold into the target. RNAdos addresses this by iteratively refining candidate sequences, selecting those that minimize the distance between predicted and target structures.

How does RNAdos work?

The algorithm starts with random RNA sequences and applies repeated rounds of inverse folding optimization using the ViennaRNA library. In each round:

  1. A candidate sequence undergoes inverse folding to find a variant more likely to match the target
  2. The resulting sequence is evaluated by computing its minimum free energy (MFE) structure
  3. If the new sequence has lower distance to the target, it replaces the current best
  4. The process terminates early if a perfect match is found (distance = 0)

This iterative approach increases the probability of finding sequences that fold correctly compared to single-attempt methods like RNAinverse. The tradeoff is longer computation time, which scales with both the number of designs requested and the number of optimization rounds.

How to use RNAdos online

ProteinIQ provides browser-based access to RNAdos with no software installation required.

Input

InputDescription
Target StructureSecondary structure in dot-bracket notation (e.g., ((((....))))...((((....)))))

The structure uses . for unpaired nucleotides and matched parentheses for base pairs. Each opening parenthesis pairs with the corresponding closing parenthesis.

Settings

SettingDescription
Number of designsIndependent design attempts to generate (1-50, default 10). More designs increase the chance of finding perfect matches.
Optimization roundsIterations per design (10-500, default 100). Higher values improve quality at the cost of runtime.
TemperatureFolding temperature in degrees Celsius (0-100, default 37). Affects thermodynamic calculations.

Results

ColumnDescription
Design #Index of this design attempt
Designed SequenceThe optimized RNA sequence
Actual StructureMFE structure when the designed sequence is folded
MFE (kcal/mol)Minimum free energy of the actual structure
DistanceBase pair distance between actual and target structures (0 = perfect match)
Matches TargetWhether the designed sequence folds exactly into the target

Successful designs have Matches Target = true and Distance = 0. Even designs with small non-zero distances may be useful as starting points for further optimization.

Interpreting results

A distance of 0 indicates the designed sequence folds into exactly the target structure at the specified temperature. Sequences with Matches Target = true are candidates for experimental validation.

Designs that do not match perfectly may still be informative. A distance of 1-2 often indicates minor structural rearrangements, while larger distances suggest the target structure may be difficult to achieve or thermodynamically unfavorable.

If no designs achieve the target, consider:

  • Simplifying the structure (shorter stems, fewer branches)
  • Checking that the structure is balanced (equal opening and closing brackets)
  • Increasing optimization rounds

Limitations

RNAdos predicts structures based on thermodynamic models that assume equilibrium conditions. Kinetically trapped states, non-canonical base pairs, and tertiary interactions are not captured.

The algorithm is stochastic, so results vary between runs. For important applications, request multiple designs and verify candidates with experimental methods.

  • RNAinverse: Single-pass inverse folding (faster, less reliable)
  • RNAfold: Predict structure from sequence (the reverse problem)
  • RNAsubopt: Enumerate alternative structures to assess design uniqueness
  • RNAeval: Calculate energy of a sequence-structure pair
  • ViennaRNA: Access all 14 ViennaRNA methods through a unified interface