Laser wavelength and sample conditioning effects on biochemical monitoring of SARS-CoV-2 VLP production upstream stage by Raman spectroscopy

Abstract

In the current scenario of biopharmaceutical industries, it is essential to monitor bioprocesses, allowing the characterization and optimization of stages and the control of critical parameters throughout the process to ensure product quality, high yield, and productivity. Through multiparameter analytical techniques, it is possible to monitor biochemical variables with quick, financially accessible measurements and accurate results. Thus, spectroscopy coupled with chemometric methods is used both in online and offline monitoring of processes, given its non-destructive, non-invasive characteristics and which obtains quantitative data of multiple biochemical variables simultaneously. Mathematical and statistical modeling methods are used to relate the spectral data, establish relationships between parameters and make forecasts possible. Thus, the present project intends to compare the predictive capacity of two correlation methods, Partial Least Squares (PLS) and Artificial Neural Network (ANN) in the off-line monitoring of the upstream process of production of SARS-CoV-2 virus-like particles (VLP). From the analysis between the models, it is also intended to identify the effects on samples without any processing (with cells) and samples of supernatants (without cells), as well as the difference in the use of different laser wavelengths in Raman spectroscopy (785 nm and 1064 nm). Finally, the models generated will be able to estimate the density of viable cells, cell viability, metabolites, nutrients, and viral titration to facilitate monitoring, optimization, and control in the next stages of development of the bioprocess up to commercial scale.