Genomic data integration systematically biases interactome mapping.

Area of Science:

• Proteomics
• Systems Biology
• Bioinformatics

Background:

• Mapping the complete protein-protein interaction network (interactome) is crucial but challenging.
• Co-migration techniques enhance interactome mapping throughput and resolution.
• Accurate identification of interacting protein pairs from large proteomic datasets remains difficult.

Purpose of the Study:

• To rigorously analyze the impact of integrating external genomic datasets on interactome mapping using co-migration data.
• To evaluate how genomic data integration affects the discovery of novel protein-protein interactions and functional coherence.
• To assess the predictive power of computationally identified novel interactions.

Main Methods:

• Analysis of numerous co-migration datasets using diverse experimental and computational approaches.
• Systematic evaluation of genomic data integration strategies in interactome recovery.
• Comparison of novel interaction predictions from co-migration data versus genomic data integration.

Main Results:

• Genomic data integration increases the functional coherence of interactome maps.
• This integration comes at the cost of reduced power to discover novel protein-protein interactions.
• Putative novel interactions identified via genomic data integration show no increased likelihood of future experimental validation compared to those from co-migration data alone.

Conclusions:

• Genomic data integration is a widespread methodology in interactome mapping with unappreciated limitations.
• The current approach may hinder the discovery of truly novel interactions.
• Re-evaluation of computational pipelines for interactome mapping is warranted.