From pull-down data to protein interaction networks and complexes with biological relevance.

Area of Science:

• Proteomics
• Systems Biology
• Bioinformatics

Background:

• High-throughput Mass Spectrometry (MS) advances enable genome-wide protein-protein interaction discovery.
• Computational inference of protein interaction networks and complexes from MS data remains challenging.
• Robust methods are needed for assessing protein interaction affinities, network representation, complex discovery, and biological relevance.

Purpose of the Study:

• To introduce a user-friendly framework for identifying protein complexes from MS pull-down data.
• To develop a computational approach for assessing protein-protein interaction affinities and constructing networks.
• To identify protein complexes and evaluate their biological relevance using graph-theoretical and statistical methods.

Main Methods:

• Assesses protein interaction affinity via co-purification pattern similarity in MS data.
• Constructs protein interaction networks using a knowledge-guided threshold selection.
• Identifies protein complexes and core components using graph theory; evaluates biological relevance statistically.

Main Results:

• The framework identified 610 protein complexes with high functional homogeneity in Saccharomyces cerevisiae.
• Achieved at least a 10% improvement in F(1)-measure compared to other methods.
• Identified potential sources of false positives (e.g., DNA-mediated co-purification) and false negatives (e.g., hydrophilic bias).

Conclusions:

• The developed framework provides an effective and robust method for protein complex identification from MS data.
• The findings highlight the importance of considering non-protein mediators and technology biases in MS-based interactomics.
• The study contributes to advancing the computational analysis of protein interaction networks and complexes.