RRIkinDP: Targeted RNA-RNA interaction kinetics

MotivationRNA-RNA interactions play essential roles in gene regulation and are controlled by both thermodynamics and kinetics. State-of-the-art tools predict thermodynamically optimal RNA-RNA interactions, but they neglect kinetic effects. While folding kinetics of single RNAs have been successfully modeled using transition systems between conformations, analogous approaches for RNA-RNA interactions quickly lead to infeasibly large systems. Novel models with controlled system size are required to overcome these limitations and enable computational analysis of RNA- RNA interaction kinetics. Such methods have the potential to improve our understanding of the governing principles of RNA-RNA interaction formation and improve target prediction tools. ResultsWe propose a targeted interaction kinetics model that focuses on a given candidate interaction structure, and further limits the state space by considering only direct paths. This allows us to describe interaction formation as a Markov process on the limited state space and to study which properties are relevant for interaction formation. By comparing experimentally confirmed sRNA-mRNA interactions in E. coli with a randomized background, we show that native interactions are indeed kinetically favored and identify key features, such as seed accessibility and folding energy barrier. Using a machine learning classifier, we identified most-informative combinations of interaction features with respect to the kinetic behavior of native RNA-RNA interactions. Our kinetics model enables the efficient computation of various features that can be used to evaluate genome-wide target predictions by kinetic criteria. Beyond immediate practical improvements, our results contribute to long-standing general questions such as the influence of initial contact site accessibility. Availability and implementationRRIkinDP is available as free software on GitHub at https://github.com/mwaldl/RRIkinDP. Contactmaria@tbi.univie.ac.at