MC-DuplexFold source code

The RNA secondary (2D) structure prediction problem involves determining the set of base pairs that form within an RNA molecule based on its sequence. A specific case of this is RNA hybridization, where two RNA strands interact to form a duplex. Thermodynamics-based RNA structure prediction methods typically use dynamic programming to compute the minimum free energy structure, relying on experimentally determined energy contributions. Through the Boltzmann distribution, these energy values can be translated into base pairing probabilities. Here, we leverage these probabilities to simulate RNA:RNA interaction dynamics. Inspired by the Ising model, we apply Gibbs sampling to model the stochastic formation and disruption of base pairs over time, ultimately deriving a consensus structure. Our program, MC-DuplexFold (mcdf), improves base-pair prediction accuracy when combined with other RNA 2D structure prediction algorithms. Benchmarking results reveal two key trends. First, approximate free energy minimization methods, such as LinearFold (Mathews Lab) and RIsearch (Gorodkin Lab), outperform exact methods like RNAcofold (Vienna Package) and DuplexFold (Mathews Lab) in structural prediction accuracy. Second, mcdf provides structural activity statistics that may enhance the modeling of miRNA primary transcripts, precursors, and target interactions, refining predictions of miRNA:mRNA duplex dynamics.