Related Experiment Videos
Co-evolutionary analysis reveals insights into protein-protein interactions
1Program in Biological and Medical Informatics, University of California, San Francisco, CA 94143, USA.
Journal of Molecular Biology
|November 8, 2002
Summary
This study introduces a computational method using evolutionary histories to predict protein-protein interactions. The approach accurately identifies known binding partners and discovers new interactions for uncharacterized proteins.
Area of Science:
- * Molecular Biology
- * Bioinformatics
- * Evolutionary Biology
Background:
- * Protein-protein interactions are fundamental to biological processes.
- * Experimental and computational methods exist to identify these interactions.
- * Computational approaches aid in inferring protein function and interactions for new genes.
Purpose of the Study:
- * To develop and extend a quantitative computational method for identifying interacting proteins.
- * To leverage correlated evolutionary histories of protein ligands and receptors.
- * To predict binding partners for proteins with unknown specificities.
Main Methods:
- * Developed a quantitative method based on correlated evolutionary histories of protein ligands and receptors.
- * Studied six diverse ligand-receptor families: syntaxin/Unc-18, GPCR/G-alpha's, TGF-beta/receptor, bacterial immunity/colicin, chemokine/receptor, and VEGF/receptor.
- * Applied the method to identify binding partners for uncharacterized proteins within the syntaxin/Unc-18 and TGF-beta/receptor families.
Main Results:
- * Achieved an average of 79% accuracy in identifying known binding partners above a defined correlation threshold.
- * Successfully predicted plausible binding partners for proteins with uncharacterized binding specificities.
- * Demonstrated that co-evolutionary analysis reduces the search space for identifying protein interactions.
Conclusions:
- * Correlated evolutionary histories effectively identify protein-protein interactions.
- * The method aids in discovering new binding partners for both characterized and uncharacterized proteins.
- * This approach can guide experimental efforts to categorize physiologically and pathologically relevant protein interactions by utilizing genomic and proteomic data.