Raman spectroscopy-based identification of nosocomial outbreaks of the clonal bacterium Escherichia coli

Abstract

DNA-based techniques are frequently used to confirm the relatedness of putative outbreak isolates. These techniques often lack the discriminatory power when analyzing closely related microbes such as E. coli. Here the value of Raman spectroscopy as a typing tool for E. coli in a clinical setting was retrospectively evaluated.

Fig. 1

Cluster analysis using the dedicated SpectraCellRA® software. Cluster analysis of the spectra was performed using the pairwise similarities (or R2-values) as a distance matrix. These R2-values are visualized in a dendrogram (left-hand side) and a correlation matrix table (right-hand side). Each horizontal line in the matrix table represents all R2-values of an isolate with all other isolates in the matrix. Strain relatedness was graphically depicted in a dendrogram plot in which each node represented the lowest correlation coefficient (or similarity) between all isolates combined in the cluster defined by this node. The horizontal axis depicts arbitrary units that are specific for this dataset and thus cannot be directly compared with other dendrogram plots obtained from different sets of isolates. By sorting these correlation coefficients based on height, those isolates with high similarity are grouped together and the color indicates the relatedness between isolates. The red clusters indicate isolates that have a similarity ≥95 %. Based on previously set thresholds [3] these isolates are classified as indistinguishable. The isolates with a similarity ≥85 to < 90 % are indicated by yellow, and those ≥90 % to < 95 % by orange. The latter two groups represent clearly distinguishable isolates that have however a high probability to be related. The white areas indicate isolates that have a similarity below 85 % and are classified as non-related ^a Epidemiologically related control isolates from semen of a single patient ^{b, c, d} Twins; * non-related control isolates