Multiscale simulation of small peptides: consistent conformational sampling in atomistic and coarse-grained models.

A bottom-up coarse-graining procedure for peptides in aqueous solution is presented, where the interactions in the coarse-grained (CG) model are determined such that the CG peptide samples conformations according to a high-resolution (atomistic) model. It is shown that important aspects of conformational sampling, such as correlated degrees of freedom (DOF) which play an important role in secondary structure formation, can be reproduced in the CG description. In some cases, microscopic structural/conformational details are lost in the coarse-graining process. We show that these "lost" properties can be recovered in a backmapping procedure which reintroduces atomistic DOF into CG structures - as long as the overall conformational sampling of the molecule is correctly represented in the CG level of resolution. Thus, it is possible to link an existing all-atom model of a biomolecular system with a CG description such that after inverse mapping one can recover structures at high resolution with the correctly sampled (according to the atomistic model) conformational properties.