Abstract
Synthesis planning is a critical process in chemical design. A number of computer programs have been developed to assist the chemists with this procedure. Most of the programs utilize combinations of computational approaches. These have been successfully applied to a number of the synthesis predictions. However, they require numerous rules to screen for potential targets, as well as to keep the system from reaching the combinatorial explosion. This results in the advanced algorithms becoming more complex and parameter-sensitive. This can be problematic, particularly in the cases in which a large number of the compounds are to be handled, because it can not only result in a lengthy computational time, but also cause some of the highly potential targets to be missed.
We developed a simpler approach for the reaction prediction using a term-rewriting grammar. The term-rewriting strategy is used to directly assign reactions to the compounds. This greatly reduces the number of rules that are usually required for these steps, and, hence, facilitates the prediction performance, while maintaining the prediction accuracy. In this dissertation, the designs of the developed algorithms and their results are first being presented, followed by a discussion of the approach’s application in the chemical design in the final chapter.
|
