Monitoring of complex industrial bioprocesses for metabolite concentrations using modern spectroscopies and machine learning: Application to gibberellic acid production

Abstract

Two rapid vibrational spectroscopic approaches (diffuse re¯ectance±absorbance Fourier transform infrared [FT-IR] and dispersive Raman spectroscopy), and one mass spectrometric method based on in vacuo Curie-point pyrolysis (PyMS), were investigated in this study. A diverse range of unprocessed, industrial fedbatch fermentation broths containing the fungus Gibberella fujikuroi producing the natural product gibberellic acid, were analyzed directly without a priori chromatographic separation. Partial least squares regression (PLSR) and arti®cial neural networks (ANNs) were applied to all of the information-rich spectra obtained by each of the methods to obtain quantitative information on the gibberellic acid titer. These estimates were of good precision, and the typical root-meansquare error for predictions of concentrations in an independent test set was <10% over a very wide titer range from 0 to 4925 ppm. However, although PLSR and ANNs are very powerful techniques they are often described as`b lack box'' methods because the information they use to construct the calibration model is largely inaccessible. Therefore, a variety of novel evolutionary computationbased methods, including genetic algorithms and genetic programming, were used to produce models that allowed the determination of those input variables that contributed most to the models formed, and to observe that these models were predominantly based on the concentration of gibberellic acid itself. This is the ®rst time that these three modern analytical spectroscopies, in combination with advanced chemometric data analysis, have been compared for their ability to analyze a real commercial bioprocess. The results demonstrate unequivocally that all methods provide very rapid and accurate estimates of the progress of industrial fermentations, and indicate that, of the three methods studied, Raman spectroscopy is the ideal bioprocess monitoring method because it can be adapted for on-line analysis.

Quantitative Analysis of the Banned Food Dye Sudan-1 Using Surface Enhanced Raman Scattering with Multivariate Chemometrics

“…Considering the fact that the response of SERS intensity with the concentration of the analyte may not necessarily be linear, we also employed two nonlinear methods: ε-support vector regression (SVR) programmed in Matlab 33 and artificial neural networks (ANNs) using an in house program. 34 For ANN analysis a multilayer perceptron with a topology of 650-8-1 (650 input Raman scatters, eight nodes in the hidden layer and a single output node for the concentration of Sudan-1) was used employing a learning rate of 0.2 and a momentum of 0.8. 35 For all three multivariate methods the performances of these three models on our data set was compared.…”

Differentiation of Micromonospora Isolates from a Coastal Sediment in Wales on the Basis of Fourier Transform Infrared Spectroscopy, 16S rRNA Sequence Analysis, and the Amplified Fragment Length Polymorphism Technique

“…The aluminum plate was then loaded onto the motorized stage of a reflectance thin-layer chromatography accessory and analyzed as detailed (10,35,43,47). The IBMcompatible personal computer used to control the IFS28 was also programmed Ϫ1 .…”

“…The resolving power of FT-IR is excellent, and it has even been shown to detect differences between yeast mutants (39) and in the metabolic footprint of microbial cells differing in the presence of single genes (25). Our established protocol for FT-IR (10,35,43,47,48) has the major advantages that it is nondestructive, reproducible, and very rapid both for a single sample (1 to 10 s) and with respect to the automated high throughput of samples in batches of 384.…”

Applications of Raman Spectroscopy in Biopharmaceutical Manufacturing: A Short Review

“…The efficacy of Raman spectroscopy for the non-invasive, in-line monitoring of different processes was first demonstrated in the 1980s and 1990s. Much of the early works focused on the production of simpler molecules, such as yeast-based fermentations for the production of ethanol, [153][154][155] and other products such as plant hormones 156 and a carotenoid antioxidant. 157 In most of these studies, the product molecule produced a reasonably strong Raman signal that had bands distinct from the (broad, less well defined) background.…”