Tumor Classification using MS Spectra Based on Deep Learning

Abstract

Deep learning models plays a significant role in bioinformatics research, such as prediction of incidence, classification of disease samples, identification and detection of tumor areas. Mass spectrometry (MS) has been widely applied to protein research due to its high throughput and sensitivity. Tumor protein mass spectrometry data has high sample dimensions and low signal-to-noise ratio, which is difficult to extract features for classification. Here, we aim to develop a new method to extract features from tumor protein mass spectrometry data and classify proteomics data using deep learning models. Our results demonstrated that the deep learning models we proposed has a good performance and may provide ideas for researchers to classify other protein mass spectral data or similar data.