Integrating Multimeric Threading With High-throughput Experiments for Structural Interactome of Escherichia coli.
Weikang Gong1, Aysam Guerler2, Chengxin Zhang2
1Department of Computational Medicine and Bioinformatics, University of Michigan, Ann Arbor, MI 48109, USA; Faculty of Environmental and Life Sciences, Beijing University of Technology, Beijing 100124, China.
Journal of Molecular Biology
|March 20, 2021
Summary
We developed Threpp, a computational pipeline to accurately predict protein-protein interactions (PPIs) and their structures. Threpp enhances genome-wide PPI network analysis and structural modeling, improving upon traditional methods.
Area of Science:
- Structural Biology
- Computational Biology
- Bioinformatics
Background:
- Genome-wide determination of protein-protein interactions (PPIs) is challenging due to high false-positive rates in high-throughput experiments (HTEs).
- Solving PPI quaternary structures is more difficult than tertiary structures using conventional techniques.
Purpose of the Study:
- To introduce Threpp, a uniform computational pipeline designed to address limitations in PPI detection and structural determination.
- To improve the accuracy and scope of genome-wide PPI network analysis and complex structure modeling.
Main Methods:
- Threpp utilizes sequence threading through a complex structure library, integrating alignment scores with HTE data via a naive Bayesian classifier.
- Quaternary complex structures are built by reassembling monomeric alignments with dimeric threading frameworks using interface-specific structural alignments.
Main Results:
- Applied to E. coli, Threpp identified 35,125 confident PPIs, a 4.5-fold increase over HTE alone.
- Constructed complex structure models for predicted PPIs, with 6,771 achieving high confidence (TM-score >0.5) and 39 showing strong consistency with experimental structures (avg. TM-score = 0.73).
Conclusions:
- Threpp significantly enhances genome-wide PPI network detection and complex structural construction.
- Threading-based homologous modeling proves effective for both PPI network discovery and structural prediction.
