A Novel Computational Approach for Identifying Essential Proteins From Multiplex Biological Networks

Area of Science:

• Systems Biology
• Bioinformatics
• Computational Biology

Background:

• Essential proteins are key to cell survival and development.
• High-throughput data enables essential protein detection via protein interaction networks (PINs).
• Existing network-based methods struggle with noisy PIN data (false positives/negatives).

Purpose of the Study:

• To develop a novel framework for identifying essential proteins by integrating multiple biological data types.
• To address the challenge of analyzing aggregated multiplex interactions for essential protein prediction.
• To improve the accuracy of essential protein identification compared to existing approaches.

Main Methods:

• Constructed a multiplex biological network (MON) integrating PINs, protein domains, and gene expression profiles.
• Extended the random walk with restart algorithm to a tensor representation for MON analysis.
• Incorporated node importance and diverse interaction types into the iterative prediction process.

Main Results:

• The MON approach was applied to identify essential proteins in two yeast PIN datasets.
• Demonstrated superior performance of MON over 11 state-of-the-art methods.
• Achieved higher accuracy based on precision-recall and jackknife curves.

Conclusions:

• The proposed MON approach effectively integrates diverse biological data for essential protein identification.
• This method overcomes limitations of traditional network-based approaches using noisy PIN data.
• MON offers a robust and accurate framework for discovering essential proteins.