SERS-CBP:A Comprehensive SERS Spectral Database ofClinically Important Bacterial Pathogens

Abstract Rapid and accurate detection of clinically important bacterial pathogens (CBP) is a public health priority. Over the past decades, the landscape of bacterial identification has experienced substantial transformation, and many diagnosis techniques for bacterial pathogens have been developed. Despite these advancements, precise identification of bacterial pathogens to inform therapeutic decisions for infectious diseases continues to pose a significant clinical hurdle, which urgently requires the development of novel diagnostic methods. As an emerging technique, surface-enhanced Raman spectroscopy (SERS), due to its high sensitivity and rapid response, the technique shows great potential in diagnosing bacterial pathogens. However, the SERS technique is currently not used in clinical settings because of the absence of a comprehensive SERS spectral database of bacterial pathogens. In this study, we manually collected 75,367 SERS spectra for 192 bacterial species belonging to 88 genera. A database web server termed SERS-CBP (http://www.sers-cbp.net/) was constructed to store all the spectral data, which can also automatically predict unknown SERS spectra submitted by anonymous users through pretrained convolutional neural network (CNN) models, which were integrated into the web server. Taken together, we constructed the first comprehensive SERS spectral database for representative bacterial pathogens commonly encountered in clinical laboratories. The SERS-CBP database holds the potential to facilitate the application of Raman spectroscopy for the identification of bacterial pathogens in clinical settings. Continuous enhancements and expansions are set to establish this database as the definitive standard for bacterial SERS spectra, solidifying its role as an essential asset for scientific and diagnostic endeavors.