ProInfer: An interpretable protein inference tool leveraging on biological networks

Area of Science:

• Proteomics
• Bioinformatics
• Computational Biology

Background:

• Protein inference is crucial in mass spectrometry (MS)-based proteomics for identifying proteins from peptide data.
• Existing methods struggle with ambiguous peptides, limiting proteome coverage and statistical confidence.

Purpose of the Study:

• To introduce ProInfer, a novel protein assembly method enhancing protein inference in MS-based proteomics.
• To improve the recovery of weakly supported proteins and increase statistical confidence using biological network information.

Main Methods:

• ProInfer leverages biological network information for protein assembly.
• It specifically addresses ambiguous peptides, which map to multiple candidate proteins.
• The method enhances statistical confidence for proteins identified by both unique and ambiguous peptides.

Main Results:

• ProInfer demonstrated superior performance in rescuing weakly supported proteins across multiple datasets (THP1, lung cancer, RAW267.4).
• It consistently inferred a higher number of true positives compared to mainstream tools like Fido, EPIFANY, and PIA.
• The tool effectively identified differentially expressed proteins, aiding functional and phenotype analysis.

Conclusions:

• ProInfer significantly improves proteome coverage and statistical confidence in protein inference.
• Its ability to utilize biological networks makes it a powerful tool for proteomics research.
• ProInfer offers enhanced capabilities for functional analysis and phenotype profiling through improved protein identification.