(532h) Fully Automated Molecular Design with Atomic Resolution for Desired Properties

The
value of fine and specialty chemicals is often determined by the specific
requirements in their physical and chemical properties. Therefore, it is most
desirable to design the structure of chemicals to meet some targeted material
properties. In the past, the design of specialty chemicals has been based
largely on experience and trial-and-error. However, recent advances in
computational chemistry and machine learning can offer a new path to this
problem. We applied the genetic algorithm (GA), based on the Darwin¡¦s theory of
evolution and natural selection, combining with simulated annealing (SA) for
this purpose. We have developed a new molecular data structure that allowed for
efficient creation of new chemical species with GA. In this presentation, we
demonstrate two successful examples where the structure of a chemical of
specified value of octanol-water partition function
and the energy gap (LUMO-HOMO) can be designed by computers. In prediction of K_ow, it consists of two parts, a robust method,
the COSMO-SAC activity coefficient model, that predicts the activity coefficient
with input of only the chemical structure, and the atom-based GA-SA. In the prediction of energy gap, we use AM1
method to calculate the energy gap for a chemical, and use GA-SA to search for
the desired value of energy gap. We will show that the target value of K_owand energy gap can be achieved
within 1% of the target with a proper set of evolution parameters (including
the size of population, the probability of selection, the rate of temperature
annealing, etc.).