DEVELOPMENT OF A MODEL OF PREDICTING THE SWEETNESS LEVEL BY USING METHODS OF GENETIC PROGRAMMING
DEVELOPMENT OF A MODEL OF PREDICTING THE SWEETNESS LEVEL BY USING METHODS OF GENETIC PROGRAMMING

Summary/Abstract: Sweeteners are natural or synthetic substances that leave a sweet feeling and have no or negligible nutritional value when compared to the range of sweetness. All over the world, there is a considerable interest in the development and synthesis of new sweeteners, and developed countries are establishing calories reduction trend, so it resorts to the use of artificial sweeteners. Due to the artificial sweetener use health risks, it is being advised to study sweeteners with a high degree of sweetness, and to reduce quantity intake. This paper follows development of quantitative models for predicting logSw (logarithm of the relative sweetness) from 132 compounds with a molecular weight of 152.146 to 804.872 and sweetness from 0 to 300000. In order to find a better link between sweetness and the structure of sweeteners, a great number of natural and artificial sweeteners of different structures will be used. The sweetness degree is correlated with quantum chemical and other molecular descriptors using evolutionary modeling, followed by analysis of the obtained models. The entire dataset is randomly divided into a training set of 26 compounds and test set consisting of 8 compounds, represented by 11 selected molecular descriptors. Values of logSw are provided by GPdotNET software. For the test set, the correlation coefficients are 0.999326. The choice of descriptors for model development is selected to be able to interpret and support the AH/B model system according to Shallenberger and Acreeu.

• Details
• Contents