How to Train a Postprocessor for Tandem Mass Spectrometry Proteomics Database Search While Maintaining Control of the...

Area of Science:

• Proteomics
• Computational Biology
• Mass Spectrometry

Background:

• Decoy-based methods are crucial for validating peptide detection in tandem mass spectrometry.
• Postprocessors like Percolator enhance statistical power by learning better scoring functions.
• Percolator can exhibit issues controlling the false discovery rate (FDR) in peptide identification.

Purpose of the Study:

• To introduce Percolator-RESET, an adaptation of the RESET meta-procedure for peptide detection.
• To ensure valid false discovery rate control in proteomics data analysis.
• To improve upon existing methods for statistical validation of peptide identification.

Main Methods:

• Percolator-RESET integrates Percolator's iterative SVM training with the RESET meta-procedure.
• The method is evaluated in both single-decoy and two-decoy modes.
• Theoretical and empirical analyses are used to validate FDR control.

Main Results:

• Percolator-RESET demonstrates valid FDR control in both single-decoy and two-decoy modes.
• The single-decoy mode reports a comparable number of discoveries to Percolator.
• The two-decoy mode offers slightly higher statistical power and reduced variability.

Conclusions:

• Percolator-RESET provides a robust solution for accurate false discovery rate control in peptide detection.
• The two-decoy mode presents an advantageous option for increased power and stability in proteomics studies.
• This method enhances the reliability of peptide identification from mass spectrometry data.