DeepMS: super-fast peptide identification using end-to-end deep learning method.

Area of Science:

• Proteomics
• Computational Biology
• Bioinformatics

Background:

• Mass spectrometry (MS) is crucial for omics analysis, especially proteomics.
• Identifying peptide sequences from MS spectra is computationally intensive and time-consuming.
• This bottleneck limits the widespread adoption of MS-based omics technologies.

Purpose of the Study:

• To develop a novel algorithm for rapid and accurate peptide identification from MS spectra.
• To overcome the computational limitations of traditional spectra identification methods.
• To enhance the accessibility and application of MS-based omics.

Main Methods:

• Developed DeepMS, a deep learning-based spectra identification algorithm.
• Utilized benchmark tests comparing six deep learning algorithms.
• Selected VGG16 as the core model for DeepMS due to its performance.

Main Results:

• DeepMS achieves super-fast, end-to-end peptide sequence identification from MS spectra with high accuracy.
• The algorithm's identification speed exceeds the MS spectra generation rate.
• DeepMS demonstrates adaptability to post-translational modifications.

Conclusions:

• DeepMS significantly overcomes the speed limitations of traditional peptide identification methods.
• The algorithm shows practical utility in microorganism detection for clinical testing.
• DeepMS has the potential to revolutionize MS-based proteomics and broaden omics technology applications.