Experimental and statistical considerations to avoid false conclusions in proteomics studies using differential...

Area of Science:

• Quantitative proteomics
• Statistical analysis in biology

Background:

• Multiple testing in proteomics can lead to an accumulation of false positives.
• Traditional reliance on p-values and significance thresholds overlooks the multiple testing effect.
• False discovery rate (FDR) quantifies false positives within significant results.

Purpose of the Study:

• To investigate bias in p-value distribution in three-dye DIGE experiments.
• To demonstrate the utility of q values for controlling FDR in proteomics.
• To highlight the importance of experimental design in statistical analysis.

Main Methods:

• Analysis of p-value distribution in three-dye DIGE experiments with a common internal standard.
• Comparison of results from three-dye versus two-dye experimental designs.
• Application and evaluation of q values in two independent proteomics studies.

Main Results:

• A bias in p-value distribution was identified in three-dye DIGE experiments due to data correlation from a common internal standard.
• This bias was absent in two-dye experiments using individual internal standards.
• In one study, 80% of significant findings using traditional methods were identified as false positives.
• Q value calculation provided control over the FDR in the second study.

Conclusions:

• Q values offer a powerful and user-friendly method for correcting multiple testing in quantitative proteomics.
• Robust experimental design, including appropriate statistical procedures, is crucial for accurate results.
• The q value approach enhances the reliability of identifying true biological changes in protein expression.