Detecting differential protein abundance by combining peptide level P-values.

Area of Science:

• Proteomics
• Mass Spectrometry
• Bioinformatics

Background:

• Label-free LC-MS bottom-up proteomics commonly infers protein abundance from peptide data.
• Existing methods often rely on summarized protein abundances or averaged peptide statistics, potentially introducing errors.
• Accurate detection of differential protein abundance is crucial for biological discovery.

Purpose of the Study:

• To develop a novel statistical method for detecting differentially abundant proteins directly at the peptide level.
• To improve the accuracy and reliability of differential proteomics analysis in label-free LC-MS experiments.
• To validate the performance of the new method against existing workflows.

Main Methods:

• Directly statistically testing peptide ionization intensities from label-free LC-MS data.
• Combining dependent P-values using the Empirical Brown's Method (EBM).
• Comparison with protein-level analysis and established workflows like MSstats using a spike-in dataset.

Main Results:

• The peptide-level approach using EBM demonstrated superior performance in differential abundance detection compared to protein-level methods on a spike-in proteomics dataset.
• The method successfully identified enriched proteins in an activity-based protein profiling dataset, showcasing its practical applicability.
• Empirical Brown's Method (EBM) effectively mitigates errors associated with protein abundance estimation.

Conclusions:

• Directly analyzing peptide ionization intensities with EBM offers a more accurate and robust method for differential protein detection in label-free LC-MS proteomics.
• This peptide-centric approach enhances the discovery power of proteomics studies.
• The developed method provides a valuable alternative to current protein-level analysis strategies.