Spectral Prediction Features as a Solution for the Search Space Size Problem in Proteogenomics

Area of Science:

• Proteomics
• Bioinformatics
• Computational Biology

Background:

• Proteogenomics aims to link genomic information with protein expression data.
• Distinguishing true peptide-spectrum matches from false positives is challenging, especially with large protein databases.
• Tandem mass spectrometry (MS/MS) is crucial for peptide identification, but accuracy can be limited.

Purpose of the Study:

• To enhance peptide identification rates and validation stringency in proteogenomics.
• To evaluate the utility of spectral intensity predictors in improving peptide-to-spectrum matching.
• To integrate features from MS2PIP and Prosit with existing scoring methods for proteogenomic analysis.

Main Methods:

• Utilized protein sequence databases generated from ribosome profiling and nanopore RNA-Seq data.
• Extracted features from MS2PIP and Prosit, tandem mass spectrometry intensity predictors.
• Combined these features with canonical scores from MaxQuant within the Percolator postprocessing tool.
• Applied the integrated approach to assess peptide identification and validation.

Main Results:

• Demonstrated that incorporating spectral intensity predictor features significantly enhances peptide identification rates.
• Showed improved validation stringency for peptide-to-spectrum matches.
• Confirmed the effectiveness of the combined approach for proteogenomic datasets.

Conclusions:

• The integration of MS2PIP and Prosit features with MaxQuant scores in Percolator offers a robust method for improving proteogenomic analyses.
• This strategy effectively addresses the challenge of distinguishing true from false peptide-spectrum matches.
• The findings support the use of spectral intensity prediction features for more confident and comprehensive proteogenomic discoveries.