A novel method for MicroRNA secondary structure prediction using a bottom-up algorithm

MicroRNAs (miRNAs) are newly discovered endogenous small non-coding RNAs (21-25nt) that are thought to regulate expression of target genes by direct interaction with mRNAs. MicroRNAs have been identified through both experimental and computational methods, and microRNA secondary structure prediction is important and essential. Generally, there are two classes of methods to predict the secondary structure of RNAs. Thermodynamics-based methods have been the dominant strategy for single-stranded RNA secondary structure prediction for many years. Recently, probabilistic-based methods have emerged to replace the free energy minimization methods for modeling RNA structures. However, the accuracies of the currently best available probabilistic-based models have yet to match those of the best thermodynamics-based methods. So this situation motivates us to develop a new prediction algorithm which will focus on microRNA structure prediction with high accuracy. A new model, nucleotide cyclic motifs (NCM), was recently proposed by Major et al. to predict RNA secondary structure. We propose and implement a novel model based on a Modified NCM (MNCM) model with a physics-based scoring strategy to tackle the problem of microRNA folding. By making use of a global optimal algorithm based on the bottom-up local optimal solutions, we implement MicroRNAfold. Our experimental results show that MicroRNAfold outperforms the current leading prediction tools.