Simple Selection Procedure to Distinguish between Static and Flexible Loops

Loops are the most variable and unorganized elements of the secondary structure of proteins. Their ability to shift their shape can play a role in the binding of small ligands, enzymatic catalysis, or protein-protein interactions. Due to the loop flexibility, the positions of their residues in solved structures show the largest B-factors, or in a worst-case scenario can be unknown. Based on the loops' movements' timeline, they can be divided into slow (static) and fast (flexible). Although most of the loops that are missing in experimental structures belong to the flexible loops group, the computational tools for loop reconstruction use a set of static loop conformations to predict the missing part of the structure and evaluate the model. We believe that these two loop types can adopt different conformations and that using scoring functions appropriate for static loops is not sufficient for flexible loops. We showed that common model evaluation methods, are insufficient in the case of flexible solvent-exposed loops. Instead, we recommend using the potential energy to evaluate such loop models. We provide a novel model selection method based on a set of geometrical parameters to distinguish between flexible and static loops without the use of molecular dynamics simulations. We have also pointed out the importance of water network and interactions with the solvent for the flexible loop modeling.