EPIFANY: A Method for Efficient High-Confidence Protein Inference.

Area of Science:

• Proteomics
• Bioinformatics
• Computational Biology

Background:

• Accurate protein inference is crucial in bottom-up proteomics, especially when dealing with shared peptides.
• Existing heuristic methods are fast but less accurate, while advanced probabilistic methods are accurate but slow for large datasets.

Purpose of the Study:

• To develop a novel protein inference method, EPIFANY, that combines accuracy with computational efficiency.
• To address the challenge of shared peptides in protein inference.

Main Methods:

• EPIFANY utilizes a loopy belief propagation algorithm integrated with convolution trees for efficient Bayesian network processing.
• The method was evaluated on the 2016 iPRG protein inference benchmark dataset.

Main Results:

• EPIFANY achieved reliable protein inference comparable to Bayesian methods but with significantly reduced runtimes.
• It was the only method to identify all true-positive proteins at a 5% protein false discovery rate (FDR) without strict prefiltering.
• Demonstrated a +10% increase in true positives and +14% in partial AUC compared to previous approaches, processing large datasets in minutes.

Conclusions:

• EPIFANY offers a robust and efficient solution for protein inference in proteomics, particularly for large-scale studies.
• Improved inference quality, especially with shared peptides, leads to more robust biological hypotheses.
• EPIFANY is available as open-source software.