Classification of deep-sea cold seep bacteria by transformer combined with Raman spectroscopy

Therefore, it is very important to conduct research on deep cold seep bacteria and the identification and screening of specific functional bacteria1. In addition, Raman spectroscopy combined with machine learning, deep learning and other classification methods can reflect the differences between different species of microorganisms, thus enabling the identification of bacteria. Deep learning-based classification methods have recently received a lot of attention because they perform well in certain classification tasks22,23,24,25. In this study, we build a Raman spectral database containing eight species of deep-sea cold seep bacteria and propose a new method for classifying Raman spectra using transformer structures. We use the commonly used classification methods and transformer structures to identify Raman spectra of deep-sea cold seep bacteria, and compare the identification results of different methods.