The idea behind the development and implementation of coarse-grained (CG) protein models is to make the simulation protocols less complex and therefore faster enough to support the studies of big protein molecules, alone or integral part of large biological systems. Being adequately parametrized, those models may support the exploration of a much wider space of conformational and chemical coordinates, than the one currently accessible for all-atom protein model.
That way of trimming the side-chains significantly reduces the number of non-bonded atom-atom pair interactions when estimating their contributions to the potential and free energy. Although our original CG model has been introduced with the simplified folding model of 1975  (LEVIT AND WARSHEL NATURE 1975), our refined model has been based on a reliable description of the electrostatic effects, which allowed us to reproduce observed proteins stabilities as well as the ability to reproduce and analyze functional properties of large biological systems.
Recognized as a central part of our research activity        the developed CG-model undergoes extensive validation and extension process. For instance, the renormalization of the model allows to obtain the effective friction by applying strong forces to both the full and reduced model, while adjusting the free energy surface and friction of the lower dimensionality model until the time dependence response of both models becomes similar. So far, our renormalization approach arguably provides the most effective way for running long-time CG-based simulations, which reliably reproduce the features of the underlining microscopic system. Our paradynamics (PD) strategy uses the CG model as a reference potential for sampling the corresponding microscopic surface.
In attempt to extend our simulation protocols to very large protein systems, we went even further in the coarse-grain application, proposing new protein model, called “C-alpha/C-beta”, where both said-chain and main chain of the protein residues are trimmed and then represented with effective atoms :
Most recently, the CG-model has been employed for predicting pathways of conformational changes that occur without the support of chemical reactions. Its role there it to support the generation and examination of samples needed for deriving the corresponding simulation trajectory, and to predict the free energy profile.
We are also conducting major studies of the ability of CG models to find the pathways between different fording states. The goal is to obtain in details at least one pathway as a sequence of intermediate conformations (transition states) and obtain their CG-energy and energy of folding. Once collected, the energy profile corresponding to each structure along the pathway may allow to estimate the barrier of the transition. The CG model effectively supports the generation of transition states at lower computational cost by removing the complexity related to the generation of side-chain conformations and their subsequent optimizations. Having that simplification implemented, we might employ "the shortest path first" method for "pushing" one conformation toward another, step by step, detecting the pathway based on a walk upon the minimum energy surface in terms of evaluated intermediate states.
Visual representation of one predicted pathway is shown in the video below:
Resolution: 1152 x 648