Modeling contaminants in AP-MS/MS experiments.

Area of Science:

• Proteomics
• Computational Biology
• Biochemistry

Background:

• Affinity purification coupled with tandem mass spectrometry (AP-MS/MS) is widely used for protein-protein interaction (PPI) identification.
• High rates of false positives in AP-MS/MS data are often caused by contaminants.
• Existing methods for contaminant identification are often insufficient or require numerous control experiments.

Purpose of the Study:

• To develop a novel Bayesian approach for identifying false-positive PPIs caused by contaminants in AP-MS/MS experiments.
• To introduce a confidence assessment algorithm, Decontaminator, for improved accuracy in PPI data analysis.
• To reduce the number of control experiments required for reliable PPI identification.

Main Methods:

• A Bayesian statistical framework was employed to model contaminants.
• A confidence assessment algorithm (Decontaminator) was developed using representative control experiments.
• The algorithm assesses the significance of Mascot scores against contaminant models to assign p-values and false discovery rates.

Main Results:

• The Decontaminator algorithm demonstrated superior performance in identifying contaminants compared to previous methods.
• The method resulted in a higher overlap with known protein-protein interaction databases.
• The Bayesian approach effectively distinguishes true interactions from contaminants in AP-MS/MS data.

Conclusions:

• The Decontaminator algorithm offers an accurate and efficient method for reducing false positives in AP-MS/MS studies.
• This approach enhances the reliability of identified protein-protein interactions.
• The method optimizes resource utilization by minimizing the requirement for control experiments.