A Refined 3-in-1 Fused Protein Similarity Measure: Application in Threshold-Free Hub Detection.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Protein similarity is crucial for bioinformatics tasks like predicting interactions and identifying disease genes.
• Existing methods for protein similarity and hub-protein detection have limitations, often requiring arbitrary thresholds.

Purpose of the Study:

• To propose an improved 3-in-1 fused protein similarity measure, FuSim-II.
• To apply FuSim-II in a novel multi-objective clustering framework for detecting hub-proteins without degree cut-offs.
• To evaluate the performance of FuSim-II and the proposed hub detection method.

Main Methods:

• Developed FuSim-II by combining weighted averages of biological knowledge from Gene Ontology (GO), Protein-Protein Interaction Networks (PPINs), and protein sequences.
• Proposed a multi-objective clustering framework using FuSim-II as the proximity measure for hub-protein detection.
• Utilized H. Sapiens and M. Musculus PPINs for experiments, employing Silhouette and Davies Bouldin (DB) indices for cluster validity.

Main Results:

• FuSim-II demonstrated improved performance over existing protein similarity measures in identifying functionally related proteins.
• The proposed hub detection framework effectively identified relevant hub-proteins without relying on degree cut-offs.
• Comparative analysis showed superior performance of FuSim-II and the hub detection method against existing algorithms.

Conclusions:

• FuSim-II is an effective protein similarity measure, outperforming existing methods.
• The integrated approach offers a robust framework for hub-protein detection in bioinformatics.
• The findings contribute to advancing protein function prediction and network analysis.