Spectral Databases

Whereas the first work in this area was realized at an intralaboratory level with few reference spectra, some manufacturers of mass spectrometers created commercial spectral databases. The MALDI Biotyper system from Bruker Daltonics (Bremen, Germany) includes an ample database of bacterial strains, mycobacteria, and fungi. A robust, standardized procedure for automated bacterial analysis, including sample preparation and data analysis, has been described (Sauer et al., 2008). The entries in the database consist of representative main spectra that have been created by replicative measurements of the corresponding reference strain. The Biotyper system has been validated in several studies, such as the identification of 1371 clinical isolates, achieving 93.2% of correct identification (Bizzini et al., 2010) and the study of 980 clinical isolates, resulting in 92.2% correct species identification (van Veen et al., 2010). Another microbial identification system based on MALDI-TOF MS, including a spectral archive, is the VITEK MS platform from bioMerieux (Marcy-l’Etoile, France). This database consists of SuperSpectra that are created by the determination of a set of biomarker peaks, representative for the corresponding genera, species, and strain. The VITEK MS platform has been validated for the identification of 1129 clinical isolates, achieving 93% of correct identification (Martiny et al., 2012). When compared to the Biotyper database, similar results were obtained. The Biotyper platform has been approved by the Food and Drug Administration (FDA) for the official identification of 40 bacterial species and the VITEK MS database for 194 species (Deak et al., 2015). Further databases are the Andromas database from Andromas SAS (Paris, France) and the MicrobeLynx bacterial identification system from Waters Corporation (Manchester, UK). Today, the Andromas database is frequently used for clinical routine analysis in Europe and includes reference data for bacteria, mycobacteria, yeasts, and fungi (Dupont et al., 2010). It has been implemented into the clinical microbiology laboratory of the Necker- Enfants Malades Hospital to identify all microorganisms isolated routinely, achieving 93% and 99% of correct species identification after single and two acquisitions, respectively (Bille et al., 2012). Likewise, the MicrobeLynx database has been successfully applied to identify clinical isolates (Rajakaruna et al., 2009).

The spectral libraries mentioned are only available commercially and require high charges for access. In this sense, the main drawback of the MALDI-TOF MS fingerprinting approach is the lack of public spectral libraries. In addition, most studies carried out are aimed at the identification of clinically relevant strains. Although many human diseases are caused by the consumption of contaminated food and the previously described commercial databases also include bacterial species that are of interest in the field of food safety, the application of these libraries to food control is not always indicated due to the lack of bacterial strains isolated from food products. For this reason, the laboratory intern database SpectraBank has been created and made publicly available (www.spectrabank.org) (Bohme, Fernandez-No et al., 2012a). The whole process for the identification of an unknown bacterial strain, including sample preparation, MALDI-TOF MS measurement and data analysis by the web tool Speclust (Alm et al., 2006) and comparison to the SpectraBank library, are explained in detail. Fig. 2.1 shows the schematic procedure of the identification process. At present, SpectraBank includes open access spectral information obtained by MALDI- TOF MS for more than 200 bacterial strains of 56 different bacterial species with interest in food safety and quality. Continuous extension of the data is intended, by adding new strains and improving the data processing, analysis’, and sharing. Recently, a further in-house spectral database has been created for Vibrio spp. (VibrioBase), since the entries in the Biotyper database were not sufficient for correct species identification of Vibrio spp. strains (Erler et al., 2015).