An integrated approach to optimization of Escherichia coli fermentations using historical data.

Using a fermentation database for Escherichia coli producing green fluorescent protein (GFP), we have implemented a novel three-step optimization method to identify the process input variables most important in modeling the fermentation, as well as the values of those critical input variables that result in an increase in the desired output. In the first step of this algorithm, we use either decision-tree analysis (DTA) or information theoretic subset selection (ITSS) as a database mining technique to identify which process input variables best classify each of the process outputs (maximum cell concentration, maximum product concentration, and productivity) monitored in the experimental fermentations. The second step of the optimization method is to train an artificial neural network (ANN) model of the process input-output data, using the critical inputs identified in the first step. Finally, a hybrid genetic algorithm (hybrid GA), which includes both gradient and stochastic search methods, is used to identify the maximum output modeled by the ANN and the values of the input conditions that result in that maximum. The results of the database mining techniques are compared, both in terms of the inputs selected and the subsequent ANN performance. For the E. coli process used in this study, we identified 6 inputs from the original 13 that resulted in an ANN that best modeled the GFP fluorescence outputs of an independent test set. Values of the six inputs that resulted in a modeled maximum fluorescence were identified by applying a hybrid GA to the ANN model developed. When these conditions were tested in laboratory fermentors, an actual maximum fluorescence of 2.16E6 AU was obtained. The previous high value of fluorescence that was observed was 1.51E6 AU. Thus, this input condition set that was suggested by implementing the proposed optimization scheme on the available historical database increased the maximum fluorescence by 55%.

Identification of critical batch operating parameters in fed-batch recombinant E. coli fermentations using decision tree analysis.

To develop a useful fermentation process model, it is first necessary to identify which batch operating parameters are critical in determining the process outcome. To identify critical processing inputs in large databases, we have explored the use of Decision Tree Analysis with the decision metrics of Gain (i.e., Shannon Entropy changes), Gain Ratio, and a multiple hypergeometric distribution. The usefulness of this approach lies in its ability to treat "categorical" variables, which are typical of archived fermentation databases, as well as "continuous" variables. In this work, we demonstrate the use of Decision Tree Analysis for the problem of optimizing recombinant green fluorescent protein production in E. coli. A database of 85 fermentations was generated to examine the effect of 15 process input parameters on final biomass yield, maximum recombinant protein concentration, and productivity. The use of Decision Tree Analysis led to a considerable reduction in the fermentation database through the identification of the significant as well as insignificant inputs. However, different decision metrics selected different inputs and different numbers of inputs to classify the data for each output.

The Effect of Solute Concentration on Equilibrium Partitioning in Polymeric Gels.

The effect of solute concentration on the equilibrium partitioning of sphere-like, colloidal solutes in stiff polymer hydrogels is examined theoretically and experimentally. The theoretical development is a statistical mechanics approach, and allows quantitative calculations to be performed to determine the concentration-dependent partition coefficient correct to first order in solute concentration at specific surface charge densities. The theory predicts that repulsive steric and/or electrostatic solute-fiber interactions exclude solute from the gel phase, but that repulsive solute-solute interactions cause partitioning into the gel to increase with increasing solute concentration. These trends are enhanced for larger solutes, increased fiber volume fractions, or stronger electrostatic repulsion. Partition coefficients have also been measured for two proteins, bovine serum albumin (BSA) and alpha-lactalbumin (ALA), in a system consisting of a salt solution and cubes of agarose hydrogel. To investigate the effect of electrostatic interactions, the experiments were performed at 0.15 M KCl and 0.01 M KCl. The theory underpredicts the strong electrostatic repulsion between BSA macromolecules at the lower ionic strength. The experimental results for ALA show the influence of an attractive interaction between the protein macromolecules, in addition to hard-sphere repulsive and electrostatic interactions. Copyright 2001 Academic Press.