Starting out, I'm using the LAMMPS and DynamO molecular dynamics packages to study coarse-graining schemes, and arrive at a suitable scheme to work with the intrinsically disordered protein n16N, which has been shown to be important in the biomineralisation of calcite. The coarse-graining scheme I use will need to be validated against an all-atom simulation.
Below you can see n16N in no particular conformation, realised in a coarse graining scheme due to Voth et al . In this bonds view rendered in VMD, green represents the alpha carbon (backbone site), light blue represents a polar site, black is an apolar site, dark blue is positive and red is negative, i.e. an acidic group.
In this scheme, each residue (amino acid) is represented by 1 backbone site and between 0 and 4 side-chain sites. The idea is to capture as much as possible of the functional groups using the different types of sites.
One of the first techniques I'll be employing is statistical temperature molecular dynamics , in which forces are scaled by the inverse of the statistical temperature T(U). The result is that the usual Boltzmann distribution for the probability of a state is replaced by a flat energy distribution. This makes sampling the protein energy landscape much easier since free energy barriers are no longer an issue.
 2010 Multiscale Coarse-Graining of the Protein Energy Landscape. PLoS Comput Biol 6(6): e1000827. doi:10.1371/journal.pcbi.1000827
 Kim J, Straub JE, Keyes T, 2006 Statistical-Temperature Monte Carlo and Molecular Dynamics Algorithms. Phys. Rev. Lett. 97(5). doi:10.1103/PhysRevLett.97.050601