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

Sperschneider, Jana; Datta, Amitava

We present DotKnot, a pseudoknot prediction method that incorporates stem-loop correlated pseudoknot energy parameters from the virtual bond model (42, 48) . The workflow is similar to the detection approach used in KnotSeeker (39) , with two main improvements. First, a secondary structure partition function calculation in Oðn 3 Þ time and Oðn 2 Þ space is the basis for finding a set of promising structure elements with high confidence (14) . This set includes heuristically derived bulge loops, internal loops and multiloops with low free energy. Second, pseudoknot candidates are assembled using the set of promising structure elements and, therefore, loop entropy parameters can readily be used for pseudoknot energy evaluations (42, 48) . There is no dynamic programming kernel for verification of pseudoknot candidates. This is a major step towards successful pseudoknot prediction as the vast majority of methods in the literature use simple approximations of pseudoknot energies. However, improving the folding model behind the algorithmic framework is clearly the key to accurate prediction (43) .