Katharina Anna Frings, Rumjhum Mukherjee, Vivien Schulze, Nils Heine, Nicolas Debener, Janina Bahnemann, Szymon Piotr Szafrański, Meike Stiesch, Katharina Doll-Nikutta, Maria Leilani Torres-Mapa, Alexander Heisterkamp
- The correct identification of different bacteria is a critical task in clinical applications and basic research especially in the oral cavity which has a complex bacterial community. Complementary to a variety of phenotyping and genotyping methods, we propose FTIR spectroscopy as a fast and non-destructive technique for accurate bacterial identification. This technique can be used to investigate the chemical makeup of a given sample and also allows for bacterial classification at strain level. In this work, we investigate the ability of ATR-FTIR spectroscopy to identify different oral bacteria from known laboratory stains as well as strains from patient-derived samples. Using this technique, six measured species could be classified with high accuracy (>97%) using chemometric models. Furthermore, the model which was only trained with laboratory strains could still correctly identify the patient-derived strains at the genus level. These results open the possibility of constructing aThe correct identification of different bacteria is a critical task in clinical applications and basic research especially in the oral cavity which has a complex bacterial community. Complementary to a variety of phenotyping and genotyping methods, we propose FTIR spectroscopy as a fast and non-destructive technique for accurate bacterial identification. This technique can be used to investigate the chemical makeup of a given sample and also allows for bacterial classification at strain level. In this work, we investigate the ability of ATR-FTIR spectroscopy to identify different oral bacteria from known laboratory stains as well as strains from patient-derived samples. Using this technique, six measured species could be classified with high accuracy (>97%) using chemometric models. Furthermore, the model which was only trained with laboratory strains could still correctly identify the patient-derived strains at the genus level. These results open the possibility of constructing a simplified tailored classification model based only on a target species and few other representative species, while still being able to distinguish the target species from a much larger number of other bacterial species for application to oral microbial communities.…
