Genetic algorithm for the design of molecules with desired properties.

摘要: The design of molecules with desired properties is still a challenge because the largely unpredictable end results. Computational methods can be used to assist and speed up this process. In particular, genetic algorithms have proved powerful tools wide range applications, e.g. in field drug development. Here, we propose new algorithm that has been tailored meet demands de novo design, i.e. efficient optimization based on small training sets are analyzed only number cycles. efficiency was demonstrated context several different applications. First, RNA were optimized respect folding energy. Second, spinglass as model system for multiletter alphabet biopolymers such peptides. Finally, feasibility computer-assisted molecular approach construction peptidic thrombin inhibitors using an iterative process 4 cycles computer-guided optimization. Synthesis experimental fitness determination 600 compounds from virtual library more than 1017 necessary achieve goal.