Multivariate analysis of Raman spectra applied to microbiology: Discrimination of microorganisms at the species level

Abstract

In this work, multivariate methods such as: principal component analysis (PCA) and partial least square-discriminant analysis (PLS-DA), were applied for the analysis and interpretation of Raman spectra, collected from microorganisms of different species. The main objective was to develop a methodology for a rapid and free of chemical-reagents discrimination and classification of microorganism at the species level. The raw Raman spectra of microorganisms were recorded in the spectral range of 2000 to 200 cm$^{ - 1}$. However, a detailed analysis of the results obtained by means of PCA, showed that the spectral region from 1700 to 1500 cm$^{ - 1}$, provides chemical and biochemical information highly correlated with the species of the microorganisms analyzed in this study, allowing a clear discrimination among species. Also, in order to evaluate the capability of multivariate methods to develop a classification rule, PLS-DA in a leave-one-out-cross-validation method (LOOCV) was used for the calibration and validation of a classification model, as a first approach. The results obtained for this method, showed an acceptable classification among the strains under study. On the other hand, taken into account the complexity of microorganisms' communities and the experimental procedures for their identification, discrimination and classification, the non-destructive and versatility of Raman spectroscopy and the capability of the multivariate methods for the analysis of spectral data, result useful tools for the classification and discrimination of this kind of samples.