Intelligent Automation Platform for Bioprocess Development

Abstract

High throughput technology has been increasingly adapted for drug screen and bioprocess development, due to the small amount of processing materials and reagents required and parallel experiment execution. It allows a wide design space to be explored in order to discover novel bioprocess solutions. Currently, the high throughput experiments for bioprocess development are implemented in a sequential fashion in which liquid handling system will perform the web lab experiment to prepare the samples; standalone analysis devices detect the data such as protein concentration; and specific software is used to realise the data analysis for process design or further experimentation. The aim of this paper is to show how the efficiency of the high throughput bioprocess development approach can be enhanced by creating an intelligent automation platform that systematically drives liquid handling system, analysis devices and data analysis to perform a closed-loop learning. The first generation prototype has been established which consists of three parts: automated devices, design algorithms and database. In order to prove the concept of prototype, both simulation and real experiments studies have been established. In this case study, the platform is used to investigate the solubility of lysozyme at various ion strengths and pH values. Tecan liquid handling system for experimentation as well as buffer preparation and a plate reader for uv absorption measurement to determine protein concentration were used as the automated devices. The simplex search algorithm and artificial neural network modelling were utilised as design algorithm to iteratively select the experiments to execute and determine the optimal design solution. An entity-relationship database with Tecan system configuration information and experimental data was established. The results demonstrate this integrated approach can implement experiments and data analysis automatically to provide specific bioprocess design solutions in a closed loop strategy at first time. It is a promising approach that may significant increase the level of lab automation to release the engineer from the labour intensive R&D activities and provides the base for sophisticated artificial intelligent learning in the future.