Deep learning-driven fragment ion series classification enables highly precise and sensitive de novo peptide...

Area of Science:

• Proteomics and Bioinformatics
• Computational Biology

Background:

• Accurate and high-throughput protein sequencing methods are lacking, limiting proteomics applications.
• Current methods struggle with unknown sequences, hindering variant calling and metaproteomics.

Purpose of the Study:

• Introduce Spectralis, a novel de novo peptide sequencing method.
• Enhance the utility of proteomics by improving peptide sequence identification.

Main Methods:

• Spectralis utilizes a convolutional neural network for spectral peak analysis.
• Fragment ion series classification and a peptide-spectrum confidence score are key innovations.
• The method analyzes tandem mass spectrometry data for de novo sequencing.

Main Results:

• Spectralis achieved over 40% sensitivity at 90% precision on known spectra.
• This performance nearly doubles the sensitivity of existing state-of-the-art methods.
• The method demonstrated superiority on unidentified spectra, enabling variant calling.

Conclusions:

• Spectralis represents a significant advancement in de novo peptide sequencing.
• The method's innovations substantially increase sensitivity in high-precision ranges.
• Spectralis paves the way for more broadly applicable proteomics.