The rational selection of purification processes for proteins: an expert system for downstream processing design

Area of Science:

• Biotechnology and biochemical engineering
• Computational biology and bioinformatics
• Process engineering

Background:

• Protein purification is critical in biotechnology, requiring complex separation strategies.
• Selecting optimal purification sequences involves combinatorial challenges, often addressed by expert systems (ES).
• Previous ES prototypes had limitations in handling complex purification operations and lacked comprehensive databases.

Purpose of the Study:

• To describe the state-of-the-art in using expert systems for rational selection of protein purification processes.
• To review the main issues in operation selection and process synthesis for protein purification.
• To present an improved ES prototype with enhanced database capabilities and a robust selection method.

Main Methods:

• Review of existing expert systems and their limitations in protein purification process synthesis.
• Development of a tree structure to represent the combinatorial nature of high-resolution purification stages.
• Implementation of an open-architecture, object-oriented ES with an improved database of protein physicochemical properties and contaminant data.

Main Results:

• The new ES prototype integrates a more realistic and accurate database, including contaminant protein titration curves.
• The system employs a tested method for selecting purification operations, improving upon previous expert-rule-based approaches.
• The enhanced ES provides users with multiple purification sequence alternatives based on extensive data.

Conclusions:

• The developed expert system offers a rational and data-driven approach to protein purification process synthesis.
• The use of object-oriented programming and an improved database enhances the accuracy and applicability of purification sequence selection.
• This advanced ES aids researchers in identifying optimal protein separation strategies through comprehensive data analysis.