EBprot

EBprot implements peptide-ratio-based probabilistic modeling to detect differential protein abundance from labeling-based and label-free quantitative proteomics data.

Key Features:

• Peptide-ratio-based analysis: Uses peptide-ratio information within a probabilistic framework that models the peptide–protein hierarchy and incorporates reproducibility evidence across multiple peptides per protein.
• Flexible semiparametric model: Employs a semiparametric model to accommodate skewed ratio distributions encountered in quantitative proteomics.
• MakeGrpData submodule: Transforms label-free peptide intensity data into peptide-ratio data and applies mixture modeling to enable group comparisons in datasets with missing peptide intensities.
• Efficient C++ implementation: Implemented in C++ to provide faster computation suitable for large-scale datasets (e.g., >100,000 peptides).
• Perseus plugin: Implemented as a plugin for the Perseus platform to integrate the analysis within Perseus-based workflows.

Scientific Applications:

• Enhanced detection of differentially expressed proteins (DEPs): Improves DEP detection by rewarding reproducible peptide-level evidence, producing improved receiver-operating characteristic curves and reduced false discovery rates.
• Versatile experimental designs: Applied to analyses including lung cancer subtype comparisons with biological replicates and time-course phosphoproteome studies of EGF-stimulated HeLa cells using multiplexed labeling.
• Simulation and spike-in studies: Demonstrates improved classification performance in simulation studies and spike-in datasets.

Methodology:

Computational methods include peptide-ratio-based probabilistic modeling of the peptide–protein hierarchy, a flexible semiparametric model for skewed ratio distributions, mixture modeling within the MakeGrpData submodule to convert label-free intensities into peptide ratios and handle missing values, and an implementation in C++.