Principal-component analysis for microbial L-phenylalanine production

Abstract

Based on experimental results of eight fed-batch fermentations using a recombinant L-phenylalanine-producing Escherichia coli strain, the applicability of principal-component analysis (PCA) for fermentation analysis was studied. Three principal components were identi®ed, representing approximately 90% of total variance. Among them, concentrations of tyrosine and acetate were identi®ed as key fermentation parameters. Their signi®cance was also con®rmed when measurement errors were taken into consideration by Monte-Carlo estimations. The error estimation approach was also used to investigate PCA suitability for the time-speci®c analysis of different fermentation phases. Relatively large principal-component score errors were calculated that limit the applicability of PCA for detailed fermentation course analysis.