SPEQ

SPEQ assesses MS/MS spectrum quality using deep learning to classify spectra for improved peptide identification against protein databases.

Key Features:

• MS/MS spectrum context: Operates on tandem mass spectrometry (MS/MS) spectra used for peptide identification by matching experimental spectra to theoretical spectra derived from protein databases.
• Deep neural network classification: Uses a deep learning model to classify spectra into high-quality and low-quality categories for downstream interpretation.
• Improved prediction accuracy: Demonstrates enhanced accuracy in predicting spectrum quality compared to existing prediction models.
• Pre-database search filtering: Filters out low-quality spectra prior to database searches to reduce computational load and focus analysis on informative spectra.
• Identification of overlooked spectra: Flags high-quality spectra that remain unidentified after initial database searches as candidates for further investigation.

Scientific Applications:

• Peptide identification in proteomics: Prioritizes spectra for peptide-spectrum matching to improve identification rates from MS/MS data.
• Protein characterization and biomarker discovery: Enhances reliability of protein-level analyses and supports discovery of proteomic biomarkers.
• Mass spectrometry data refinement: Improves downstream analyses relevant to disease diagnostics, drug development, and biological pathway elucidation by enriching for high-quality spectra.

Methodology:

A deep neural network is trained on labeled spectra-quality datasets to classify spectra as high- or low-quality, perform pre-database-search filtering of low-quality spectra, and is evaluated by comparison with existing prediction models.