How side chains are modeled

Amino acid side chains tend to assume one of a discrete number of "rotamer" conformations. The rotamers used in CLC Drug Discovery Workbench have been calculated from a non-redundant set of high-resolution crystal structures.

Side chains are modeled by their lowest energy rotamer. In the Protein Optimizer this is found by trying each possible rotamer in turn. For creating homology model structures with the Find and Model Structure tool, a heat bath Monte Carlo simulated annealing algorithm is used, similar to the OPUS-Rota method [Lu et al., 2008]. The algorithm consists of approximately 100 cycles of simulation. In a single cycle, rotamers are selected for each side chain with a probability according to their energy. As the simulation proceeds, the selection increasingly favors the rotamers with the lowest energy, and the algorithm converges.

A local minimization of the modeled side chains is then carried out, to reduce unfavorable interactions with the surroundings.

Calculating the energy of a side chain rotamer

The total energy is composed of several terms:

Note: For manual mutations, the atom interaction potential considers only interactions within the mutated protein chain. In the case where side chains are modeled with the Find and Model Structure tool, the interactions with all molecules in the template PDB file (except water) are considered.

Local minimization of side chain

After applying a side chain rotamer from the library to the backbone, a local minimization may be carried out for rotations around single bonds in the side chain.

The potential to minimize with respect to bond rotation is composed of the following terms: