MD simulations offer the opportunity to dynamically study complex biological and material processes beyond the scope of experimental methods. Among the biological processes, signalling pathways, enzyme regulation, protein folding and interactions, membrane composition, and transmembrane transport are of special interest. Post-translational modifications (PTMs) describe the attachment of functional groups to biomolecules after their synthesis. PTMs are abundant in both the proteome and epigenome so that their dysregulation is involved in the development of cancer and neurodegenerative diseases and plays an important role in aging. Moreover, lately, the scientific community has been becoming aware of the importance of correct lipid composition of membranes and the role of protonation in various cellular processes. However, to adequately represent physiological conditions in empirically based simulations as MD, atomic and molecular interactions have to be fine-tuned and parametrized. Coarse-grained (CG) MD simulations pave the way to even larger systems and time scales by combining several atoms into beads. Thus larger simulation systems and longer periods of time can be studied at molecular resolution. We utilize CG simulations for example to study association of transmembrane proteins in native-like membrane models or to describe mobility of reactants and products through soft porous crystals carrying organocatalysts. In order to access atomistic details of CG structures, we are developing a machine-learning based method to convert the resolution back to atomistic all-atom (AA). Such resolution-conversion techniques are not restricted to biological systems but provide a feedback control enabling a quality check of the newly established coarse-grained representation by comparison against experimental and other simulational data.