Essential proteins are critical for cell regulation, reproduction and metabolism, with their absence leading to the cessation of cell replication or cell death. Therefore, identifying essential proteins is vital for the developing antibiotics, new therapies and targeted drugs. However, current computational methods rarely or insufficiently consider the impact of features importance and the noisy data in protein-protein interaction network on recognition accuracy. To address these issues, we propose a essential protein identification method based on a hypergraph and Hierarchical Advantage Suppression Model, HGMO. Firstly, based on the principles of co-expression and co-localization, a hypergraph network is constructed using protein-protein interaction network, gene expression data, and subcellular localization data. From this network, a new feature score, the Topological Significance (TS) score, is extracted. Then, we extract subcellular localization scores and orthologous scores from the subcellular localization data and orthologous data, respectively. Finally, we uses a multi-omics data integration model, Hierarchical Advantage Suppression Model, for feature fusion, thus proposing the HGMO method. Experimental results show that HGMO consistently outperforms other methods on three S.cerevisiae datasets. Moreover, TS demonstrates a higher identification rate than other centrality methods.

• PMID: 39952483
• DOI full text
• PubMed
• PubTator

Reference Type

Journal Article

Authors

Hu Z, Kui X, Liu C, Zou Z, Li Q, Liao S, Zou B

Primary Lit For

Additional Lit For

Review For