A stem weight maximization scheme for RNA pseudoknot prediction

An RNA pseudoknot prediction algorithm based on maximization of stem weight is proposed. The algorithm involves three stem searching function that is section search, cross search and knot search which search for base pairing in different regions. Section search identifies simple RNA stems and loops without crossing. Cross search will search for inner base pairing without crossing of stems found in section search. Finally, knot search which search for potential pseudoknot structures by searching for base pairing between hairpin loop and other unpaired regions. The resulting secondary structure is represented in dot-bracket representation and could be visualized with VARNA [1]. A total of 110 RNA secondary structures downloaded from FRABASE [2] are compared with structures generated by the proposed algorithm. Structure comparison is carried out by calculating the specificity (SP) and sensitivity (SN) scores between the predicted structure to the experimentally determined structure from FRABASE.