Selection of Features with Consistent Profiles Improves Relative Protein Quantification in Mass Spectrometry...

Area of Science:

• Proteomics
• Mass Spectrometry
• Bioinformatics

Background:

• Bottom-up mass spectrometry proteomics relies on data-dependent acquisition (DDA), data-independent acquisition (DIA), or selected reaction monitoring (SRM) for relative protein quantification.
• Current workflows summarize protein abundances using all spectral features, which can be compromised by inconsistent feature patterns due to technical or biological factors.
• Such inconsistencies can undermine protein-level summaries and downstream conclusions in proteomics studies.

Purpose of the Study:

• To develop and evaluate a statistical approach for automatically detecting spectral features with inconsistent patterns in mass spectrometry-based proteomics data.
• To enable separate investigation and potential removal of problematic spectral features to improve data quality.
• To enhance the accuracy, sensitivity, specificity, and reproducibility of protein quantification and differential abundance analysis.

Main Methods:

• A novel statistical approach was developed to identify spectral features exhibiting patterns inconsistent with the overall protein profile.
• The method was evaluated using benchmark-controlled mixtures and biological samples across DDA, DIA, and SRM data acquisition workflows.
• The approach is implemented as an option within the open-source R-based software MSstats.

Main Results:

• The proposed statistical approach effectively detects spectral features with inconsistent patterns across various mass spectrometry data types.
• Evaluation on controlled mixtures and biological data demonstrated the method's ability to facilitate and complement manual data curation.
• The approach significantly improved estimation accuracy, sensitivity, and specificity for detecting differentially abundant proteins.

Conclusions:

• The developed statistical method provides a robust tool for identifying and managing problematic spectral features in mass spectrometry proteomics.
• This approach enhances the reliability and reproducibility of protein quantification and differential expression analysis.
• The integration into MSstats makes this advanced data quality control readily accessible to the research community.