An evolutionary algorithm for stochastic context-free grammar design, with applications to RNA secondary structure prediction

Stochastic Context-Free Grammars (SCFGs) have been used widely in modelling RNA secondary structure. They were motivated by the use of Hidden Markov Models (HMMs) in protein modelling (Krogh et al., (1993)). What was lacking in HMMs though, was the ability to model long range interactions which are necessary to provide an effective model for RNA secondary structure. Thus, SCFGs, as generalisations of HMMs which can model long range interactions, were used to model RNA secondary structure (Sakakibara et al. (1994), Lefebvre (1996)).