PolySTest: Robust Statistical Testing of Proteomics Data with Missing Values Improves Detection of Biologically...

Area of Science:

• Proteomics
• Bioinformatics
• Statistical Analysis

Background:

• Statistical testing is crucial for identifying differentially regulated proteins in large-scale quantitative proteomics using mass spectrometry.
• Missing values and experimental variations complicate robust statistical analysis, impacting confidence scores.
• Choosing appropriate statistical methods is challenging due to diverse experimental strategies and informatics tools.

Purpose of the Study:

• To introduce PolySTest, a user-friendly web service for statistical testing, data browsing, and visualization in proteomics.
• To present Miss test, a novel method for simultaneously testing missingness and feature abundance.
• To enhance the confidence and sensitivity of large-scale proteomics data analysis.

Main Methods:

• Development of PolySTest, a web service integrating statistical testing, data browsing, and visualization.
• Introduction of Miss test, a statistical method to handle missing data and feature abundance.
• Validation using artificial and experimental proteomics datasets with known ground truth.

Main Results:

• PolySTest with Miss test demonstrated higher confidence and sensitivity in analyzing proteomics data.
• The method successfully rescued 10-20% additional proteins in molecular networks relevant to muscle differentiation.
• PolySTest provides accurate confidence scores irrespective of instrument platform, protocol, or software.

Conclusions:

• PolySTest and Miss test are valuable tools for improving the coverage and depth of large-scale proteomics experiments.
• The approach enhances the identification of differentially regulated proteins, particularly in the presence of missing data.
• PolySTest offers a robust solution for statistical testing in mass spectrometry-based proteomics.