Raman spectroscopy is a rapid analysis method of biological samples without labeling and 
destruction. At present, the commonly used Raman spectrum classification models include CNN, 
RNN, etc. The transformer has not been used for Raman spectrum identification. This paper introduces 
a new method of transformer combined with Raman spectroscopy to identify deep-sea cold seep 
microorganisms at the single-cell level. We collected the Raman spectra of eight cold seep bacteria, 
each of which has at least 500 spectra for the training of transformer model. We compare the 
transformer classification model with other deep learning classification models. The experimental 
results show that this method can improve the accuracy of microbial classification. Our average 
isolation level accuracy is more than 97%.