Exploring bacterial phenotypic diversity using factorial design and FTIR multivariate fingerprinting. (15th January 2014)

• Record Type: Journal Article
• Authors: Wehrli, Patrick M.; Lindberg, Erika; Svensson, Olof; Sparén, Anders; Josefson, Mats; Dunstan, R. Hugh; Wold, Agnes E.; Gottfries, Johan; Åberg, Magnus
• Is Part Of: Journal of chemometrics. Volume 28:Number 8(2014:Aug.)
• Page Start: 681
• Page End: 686
• Publication Date: 2014-01-15
• Subjects: Chemistry -- Mathematics -- Periodicals; Chemistry -- Statistical methods -- Periodicals; 542.85
• DOI: 10.1002/cem.2588
• Languages: English
• ISSNs: 0886-9383

Abstract

<abstract abstract-type="main"> <title> <x xml:space="preserve">Abstract</x> </title> <p>Transmission Fourier transform infrared (FTIR) spectra were obtained from cultures of <italic>Staphylococcus aureus</italic> that were grown under sets of various environmental conditions enclosing ranges of potential conditions in wound sites. The primary aim of this study was to determine whether bacterial phenotypic diversity could be characterized by FTIR spectroscopy analyses of cultures of <italic>S</italic>. <italic>aureus</italic> grown under variable sets of environmental conditions. Designing experiments with full factorial design has shown to be a powerful way to explore an expectedly large array of phenotypic diversity of <italic>S</italic>. <italic>aureus</italic>. Various combinations of environmental conditions caused the bacteria to adapt their phenotype, which was assessed via FTIR spectral fingerprinting. Transmission FTIR spectroscopy was found to be superior to other vibrational spectroscopy techniques for this purpose because of its high sensitivity, reproducibility, and rapidity of analysis. The sample preparation presented was fundamental for reproducible results. Different spectral preprocessing methods were compared in combination with scaling methods to obtain distinguishable phenotypes in principal component analysis (PCA) models. Spectral preprocessing with combined filters, including standard normal variate transformation, Savitzky–Golay smoothing, and use of the first derivative in a PCA model with unit variance scaling, gave the most optimal clustering for the data in this study. Spectra obtained from each treatment group were found to have a unique FTIR profile with good reproducibility, and thus PCA resulted in full separation between all groups on three principal components. In conclusion, transmission FTIR spectroscopy in conjunction with design of experiment, and multivariate analysis are powerful tools to investigate phenotypic diversity of <italic>S</italic>. <italic>aureus</italic>. Copyright © 2014 John Wiley & Sons, Ltd.</p> </abstract>