Identifying Human Essential Genes by Network Embedding Protein-Protein Interaction Network.

2019 
Essential genes play an indispensable role in cell viability and fertility. Identifying human essential genes helps us to study the functions of human genes, but also provides a way for finding potential targets for cancer and other diseases. Recently, with the publishing of human essential gene data and the availability of a large amount of biological data, some computational methods have been proposed to predict human essential genes based on genes’ DNA sequence or their topological properties in the protein-protein interaction (PPI) network. However, there is still some room to improve the prediction accuracy. In this work, we propose a novel supervised method to predict human essential genes by network embedding protein-protein interaction network. Our method extracts the features of the genes in network by mapping them to a latent space of features that maximally preserves the relationships between the genes and their network neighborhoods. After that, the features are input into a SVM classifier to predict human essential genes. Two human PPI networks are employed to evaluate the effectiveness of our method. The prediction results show that our method outperforms the method that only uses genes’ sequence information, but also is obviously superior to the method utilizing genes’ centrality properties in the network as input features.
    • Correction
    • Source
    • Cite
    • Save
    • Machine Reading By IdeaReader
    41
    References
    2
    Citations
    NaN
    KQI
    []