Comparison of coarse-grained and all-atom representations by explicit energy landscape explorations

Computational studies of large molecular systems are often hindered by resource constraints, such as the available computational time. A common approach to reduce the computational cost is to use a coarse-grained description instead of an all-atom representation. However, such a simpli cation requires careful consideration of the coarse-graining scheme to identify potential artefacts introduced and the limitations of the model. In this contribution, we use the computational energy landscape framework to explicitly explore the energy landscapes for a coarse-grained (HiRE-RNA) and an all-atom potential (AMBER) for an example system, the Aquifex aeolicus tmRNA pseudoknot PK1. The method provides insight into structural, thermodynamic and kinetic properties within a common framework, and allows for a comparison of a variety of commonly computed observables, demonstrating the usefulness of this approach. For the speci c case study, we observe that both potentials exhibit a number of common features, highlighting that the coarse-grained model captures essential physical features of the system. Nonetheless, we observe shortcomings, and we demonstrate how our approach allows us to improve the model based on the insight obtained from the computational modelling.