Abstract:
Coronavirus disease 2019 (COVID-19) still provides global public health issues although several vaccines and antiviral agents have been developed. Some patients experience severe conditions needed medical intensive care, and some are dead due to the failure of treatments. Therefore, identifying the key genes and underlying molecular mechanisms is necessary to discover precisely targeted drugs. Analysis of protein-protein interaction (PPI) networks provides invaluable information to find disease mechanisms and effective alternative drugs. Hence, PPI network analysis based on leukocyte transcriptomic profiles of severe COVID-19 collected from Gene Expression Omnibus (GEO) DataSets was proposed for this study. A network diffusion method called Laplacian heat diffusion (LHD) algorithm was performed to construct an immune-related PPI network (IPIN). Furthermore, several network centrality measurements can identify 23 key genes from the IPIN. Subsequently, drug-gene interaction networks were constructed using database searching based on the key genes. There were 5 candidate drugs having the potential effect of interacting with the key genes. To find additional key genes and candidate drugs, two different leukocyte transcriptomic datasets were combined for the common PPI network construction. Centrality measurement and survival analysis were used to find and validate the further key genes. The analysis revealed 4 common key genes. The drug-gene interaction and molecular docking technique provided 2 further candidate drugs that interacted with the key genes. Additionally, miRNA-mRNA regulatory networks were built based on the PPI network to recognize 5 novel biomarkers for severe COVID-19 prediction. In conclusion, PPI network analysis can discover candidate biomarkers and drugs to predict and treat severe COVID-19 patients.
