Computer-Assisted Organic Synthesis. Addition to Carbonyl Carbons and Aromatic Substitution
This thesis concerns the computer-assisted modelling of additions to ketones and aromatic substitution, and how this modelling can be used within an existing program for synthesis planning. I include a review of existing computer-assisted organic synthesis (CAOS) programs, and discuss how some of these programs deal with knowledge representation.
The diastereoselectivities have been modelled for the addition to ketones. The steric requirements for two hypothetical transition states are calculated and the diastereoselectivity of the reaction is calculated in a very fast way (less than one second). For 399 reactions involving 223 substrates and eight reagents, diastereoselectivity data are taken from the literature and fitted to the model. The quotient between the calculated and the reported selectivity is taken as the minimization criterion. I found that steric interactions cannot alone explain the selectivity; it is necessary to include a correction term that includes the effects of torsional strain between the incoming reagent and the b-substituents of the substrate. A mean deviation of 2.17 between the literature and calculated selectivities is found, a fully satisfactory value for the purpose.
For the aromatic substitution reactions, a fast calculational scheme is derived for calculating the regioselectivity. From the literature, 976 data points from 176 substrates with 22 reagents, representing approximately 400 different positions, were correlated within the model with a correlation factor, r = 0.905. Similar models have previously been correlated with much fewer data points, and have thus given somewhat higher correlation factors. The diverse types of substrates that can be treated in this method makes it well suited for the inclusion into an automated search procedure.
aromatic substitution reaction