A dynamic noise level algorithm for spectral screening of peptide MS/MS spectra

Area of Science:

• Proteomics
• Mass Spectrometry
• Bioinformatics

Background:

• High-throughput shotgun proteomics generates numerous spectra of non-peptide ions or low quality.
• These spectra can lead to false positives or no identifications in database searches.
• Filtering these spectra enhances search accuracy and reduces computational load.

Purpose of the Study:

• Develop and evaluate an algorithm to filter low-quality tandem mass spectra from shotgun proteomic data.
• Improve the efficiency and accuracy of peptide identification in proteomics studies.

Main Methods:

• A novel algorithm dynamically determines noise levels for individual spectra.
• Spectra are filtered based on a minimum number of signal peaks with a signal-to-noise ratio of 2.
• The algorithm was tested on 23 datasets comprising 62,117 spectra.

Main Results:

• The spectral screening effectively removed 89.0% of spectra without peptide matches using MassMatrix.
• Only 6.0% of true positive peptide matches were lost during screening.
• The algorithm demonstrated high efficacy across multiple search engines (Mascot, OMSSA, X!Tandem), removing 75.93%-91.00% of unidentified spectra with minimal loss of true positives (3.59%-9.40%).

Conclusions:

• The developed algorithm is highly effective in removing unidentifiable tandem mass spectra.
• This spectral filtering significantly improves the quality of proteomics data analysis.
• The method offers a valuable tool for enhancing the accuracy and efficiency of shotgun proteomics.