DotKnot: pseudoknot prediction using the probability dot plot under a refined energy model

Sperschneider, Jana; Datta, Amitava

On the first level, two crossing stems are combined to form a core H-type pseudoknot (Figure 4) . Core H-type pseudoknots are the building blocks for more complex pseudoknots. The pseudoknot stems can either be regular stems taken from the dictionary D s or interrupted stems from the dictionary D L s ( Figure 5 ). Here, we only allow pseudoknots with at most one interrupted stem because for more complex and less rigid pseudoknots we would have to employ a highly assumptive energy model. Certain loop length restrictions are applied because more meaningful results can be expected from prediction of shorter and well-studied pseudoknots. Loop L 1 and L 3 are both required to have at least 1 and 2 nt, respectively. Interhelix loop L 2 can have a size of 0 nt. All three loops are restricted to a maximum length, as there are no reliable energy parameters for very long pseudoknots. Loops L 1 and L 3 can have a length of up to 100 nt, whereas interhelix loop L 2 is restricted to a maximum length of 50 nt. During construction of the pseudoknot candidates, we discovered that crossing stems may compete for a base pair. This leads to an overlap at loop L 2 . In such a case, one of the stems is truncated according to certain rules (Supplementary Figure S2 and Supplementary Algorithm 2).