DiffSLC: A graph centrality method to detect essential proteins of a protein-protein interaction network

Area of Science:

• Systems biology
• Bioinformatics
• Network analysis

Background:

• Identifying central genes and proteins is crucial for understanding biomolecular networks and predicting essentiality.
• Existing network centrality measures may not fully capture the complexity of biological systems.

Purpose of the Study:

• To introduce and evaluate the DiffSLC centrality measure for predicting essential genes and proteins.
• To improve the accuracy of essentiality prediction by integrating protein-protein interaction networks with gene co-expression data.

Main Methods:

• Developed DiffSLC, a novel centrality measure combining protein interaction counts, gene co-expression values, eigenvector centrality, and edge clustering coefficients.
• Utilized Saccharomyces cerevisiae protein-protein interaction networks and gene expression data for validation.
• Compared DiffSLC against existing centrality measures on seven different networks.

Main Results:

• DiffSLC demonstrated superior performance in identifying essential proteins compared to other centrality measures.
• The method achieved a higher area under the ROC curve, indicating improved prediction accuracy.
• Analysis showed the impact of including/excluding gene co-expression data and using different co-expression/gene expression datasets.

Conclusions:

• DiffSLC is a robust and effective centrality measure for identifying essential genes and proteins in biological networks.
• The integration of gene co-expression data significantly enhances the prediction of essentiality.
• DiffSLC offers a valuable alternative for essential gene detection, adhering to the centrality-lethality principle.