Bayesian Analysis of iTRAQ Data with Nonrandom Missingness: Identification of Differentially Expressed Proteins

Area of Science:

• Proteomics
• Bioinformatics
• Statistical Modeling

Background:

• Isobaric Tags for Relative and Absolute Quantitation (iTRAQ) enables simultaneous protein quantitation across multiple samples.
• Accurate inference of relative protein expression is crucial for identifying differentially expressed proteins.
• Existing methods may have limitations in handling complex iTRAQ data, including missing values.

Purpose of the Study:

• To develop a Bayesian hierarchical model for inferring relative protein expression levels from iTRAQ data.
• To identify differentially expressed proteins using a robust statistical framework.
• To account for peptide-specific effects and non-random missingness in the iTRAQ data.

Main Methods:

• A Bayesian hierarchical model was developed, treating protein expression and peptide-specific effects as random effects.
• A logistic regression model was incorporated to handle non-random missing peptide data based on protein expression levels.
• Markov chain Monte Carlo (MCMC) methods were employed for parameter inference, including relative expression levels.

Main Results:

• Simulation studies indicated that MCMC-based estimates of relative protein expression exhibit reduced bias compared to ANOVA or fold-change methods.
• The proposed model effectively handles peptide-specific variations and missing data patterns.
• The method was successfully applied to an iTRAQ dataset investigating the role of Caveolae in cardiovascular function.

Conclusions:

• The Bayesian hierarchical model provides a statistically robust approach for analyzing iTRAQ data.
• This method enhances the accuracy of differential protein expression identification.
• The findings contribute to a better understanding of protein expression in complex biological systems, such as cardiovascular function.