Quantitative structure-enantioselectivity relationships using neural networks. Bioconversion of carbonyl compounds using baker's yeast

Abstract

Quantitative structure-enantioselectivity relationships were studied for the reduction of a set of 73 carbonyl compounds with bakers yeast. The established model, using a neural network, allowed the prediction of the reduction selectivity (% S enantiomer) with success. The correlation coefficient between the observed and calculated %S was 0.99. The model was also used to predict the enantioselectivity of the reduction of adiketones using bakers yeast and different microorganisms.