EVOLVE: An Evolutionary Toolbox for the Design of Peptides and Proteins

We report the development and application of a versatile evolutionary algorithm toolbox (EVOLVE) targeted at the engineering and optimisation of peptides and proteins via atomistic simulation. As a benchmark study, we have applied the singleobjective genetic algorithm within EVOLVE to a prototypical 20 amino acid long alpha-helix, ALA6-X8-ALA6, where the sequence of the central eight amino acids (X) has been optimised in di erent chemical environments. The  tness function driving the optimisation is based on a molecular mechanics evaluation of the energy of a given sequence with respect to a reference homo-ALA helix, coupled with implicit solvent corrections. The simulations performed show EVOLVE consistently converges to low-energy solutions in a vast search space of 177^8 = ca. 10^17 possible sequences and side chain conformations, and relatively few iterations are required to reach these near-optimal structures. In all environments, the identi ed optimal sequences of the alpha-helices are considerably more stable than all possible homo-helices, and thus go beyond what is accessible via exfoliative enumeration.