An Expert System for the selection and synthesis of multistep protein separation processes

A hybrid Expert System that combines expert rules and mathematical correlations to manipulate databases for selecting the sequence of operations for the downstream purification of proteins was built. This paper describes how the physicochemical data on the protein product and the other proteins present (contaminants) was used to select a sequence of operations with a minimum number of steps to achieve a defined level of purity. An algorithm was developed to model the amount of each protein contaminat eliminated after each step. The system was tested in a practical example and it was found that process synthesis was sufficiently sensitive to important changes in the physicochemical parameters of the product and rather robust to small variations or errors in the measurement of such parameters (e.g. 10%).