It has been proven that general pseudoknot prediction in energy based models is NP-complete. Therefore, available RNA folding algorithms restrict themselves to certain classes of pseudoknots for efficiency reasons. One such class comprises the canonical simple recursive pseudoknots.

A simple pseudoknot has two helices conflicting the nesting convention and three loops intervening the pseudoknot stems. Within the loops, any other secondary structure, including further pseudoknots, is allowed, as long as they do not interact with bases outside of the loop. This makes the pseudoknot recursive. Finally, the canonizations requires, that pseudoknot helices must have maximal extent. In other words, possibly paired bases at the interior ends of the pseudoknot stems have to be paired. See Figure D for an example.

The program pknotsRG is an energy based folding algorithm that extends the standard folding algorithms (like RNAfold) by the class of canonical simple recursive pseudoknots. It computes the minimal free energy structure for a sequence, displayed as a dot bracket string. For better readability, base pairs involved in a pseudoknot are denoted with square brackets for the first stem and curly brackets for the second stem.

When using pknotsRG always remember: Due to the complexity of pseudoknot folding the run time of the program can be about a factor of the length of the sequence larger than e.g. RNAfold.

Many viruses use a pseudoknot, positioned downstream of a characteristic heptanucleotide slippery sequence, for efficient ribosomal frameshifting. In the proposed model the pseudoknot causes the ribosome to pause. While paused, the ribosome slips back one nucleotide on the slippery site and continues translation in the -1 reading frame. In the exercise we will fold two such knots.

When the pseudoknot is dominated by an unknotted structure, pknotsRG can compute the best secondary structure that contains at least one pseudoknot somewhere in the sequence.

If the user is not interested in the global folding of the molecule, but only wants promising pseudoknot candidates, the local mode of pknotsRG is the optimal choice. In the local mode pknotsRG computes the pseudoknot with the best energy to length ratio. The rest of the sequence remains unfolded.