QC metrics from CPTAC raw LC-MS/MS data interpreted through multivariate statistics.

Area of Science:

• Proteomics
• Analytical Chemistry
• Biostatistics

Background:

• Shotgun proteomics experiments involve complex processes prone to variability.
• Current quality metrics often rely on MS/MS identifications, which can be limiting.

Purpose of the Study:

• To develop and validate identification-independent quality metrics for LC-MS/MS data.
• To establish a multivariate statistical framework for assessing and visualizing variability in proteomics experiments.

Main Methods:

• Utilized QuaMeter software for identification-independent quality metrics.
• Applied a multivariate statistical toolkit including principal components analysis, factor analysis, and nested ANOVA.
• Analyzed data from Clinical Proteomics Technology Assessment for Cancer (CPTAC) Studies 1 and 5.

Main Results:

• Identification-independent metrics differentiated sites and run times, revealing performance outliers.
• Nested ANOVA demonstrated the impact of mass spectrometer and run time on metrics.
• Even with Standard Operating Procedures (SOPs), instrument-dependent variability persists, though within-site variability is reduced.
• Quality metrics predicted identification sensitivity.

Conclusions:

• A robust multivariate framework using identification-independent metrics effectively assesses variability in shotgun proteomics.
• This approach enhances the understanding of mass spectrometry performance and data quality.
• The tools facilitate improved quality control and data interpretation in large-scale proteomics studies.