Linear discriminant analysis-based estimation of the false discovery rate for phosphopeptide identifications

Area of Science:

• Proteomics
• Biochemistry
• Bioinformatics

Background:

• Liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS) enables large-scale phosphopeptide characterization.
• High-throughput, confident phosphopeptide identification and false discovery rate (FDR) estimation remain challenging.
• Accurate FDR estimation is crucial for reliable interpretation of phosphoproteomics data.

Purpose of the Study:

• To develop and validate a data analysis pipeline for high-throughput, confident phosphopeptide identification.
• To rigorously estimate the false discovery rate (FDR) for phosphopeptide identifications.
• To improve the reliability of large-scale phosphoproteomics studies.

Main Methods:

• Reanalysis of phosphopeptide identifications with ambiguous phosphate assignments to determine optimal phosphorylation site localization.
• Application of an expectation maximization algorithm to estimate peptide score distributions.
• Linear discriminant analysis to combine peptide scores (e.g., from SEQUEST) into a discriminant score for FDR estimation.

Main Results:

• A novel data analysis pipeline was established for phosphopeptide identification and FDR estimation.
• The pipeline successfully processed data from irradiated human skin fibroblasts, yielding robust FDR estimates.
• The developed method enhances confidence in phosphopeptide identifications from large datasets.

Conclusions:

• The described data analysis pipeline effectively addresses challenges in high-throughput phosphopeptide identification and FDR estimation.
• This approach provides a robust method for analyzing complex phosphoproteomics data.
• The associated Phosphopeptide FDR Estimator software is available for public use.