The identification of DEGs
In this study, 1093 DEGs were identified, of which 721 were upregulated, and 372 were downregulated, including 196 co-expressed upregulated genes and 66 co-expressed downregulated genes, (Table S1) which are roughly shown in the Venn diagram (Fig. 2a,b). Volcano plots (41, 42) (Fig. 3, a-d) and expression heatmap (Fig. 4, a-d) for all profile datasets used for data extraction and analysis were constructed. The red lines in the volcano plots represent the threshold values of the identification of DEGs.
Functional Enrichment analysis
To identify the pathways which had the most biologically and statistically significant involvement with the genes identified, upregulated and downregulated DEGs were submitted into DAVID for GO and KEGG pathway analysis. For all co-DEGs, they are mainly concentrated in cyclooxygenase P450 pathway, cell membrane, iron-binding, and metabolic pathway (Fig. 5, a-d). The upregulated co-DEGs are enriched primarily on cyclooxygenase P450 pathway, cell membrane, and oxidoreductase activity, which binds or reduces molecular oxygen and metabolic pathways by acting on paired donors (Fig. 6, a-d). As for the downregulated co-DEGs, the functional enrichment analysis reveals the results that cell division, spindle, protein binding, and cell cycle. (Fig. 7, a-d) For module 1, the genes involved in this were mainly enriched in cell division, nucleus, protein binding, and cell cycle. (Fig. 8, a-d) For module 2, the genes involved in this were primarily enriched in epoxygenase P450 pathway, organelle membrane, arachidonic acid epoxygenase activity, and drug metabolism-cytochrome P450 (Fig. 9, a-d). For module 3, the genes involved in this were primarily enriched in xenobiotic metabolic process, organelle membrane, oxygen binding, and metabolic pathways (Fig. 10, a-d). For the top 10 hub genes of the whole network, they were primarily enriched in cell division, nucleoplasm, cyclin-dependent protein serine/threonine kinase activity, and p53 signaling pathway (Fig. 11, a-d).
PPI network construction and module analysis
Interactions between the identified DEGs were elucidated by the PPI network. (Fig. 12, a-c) In total, there were 226 nodes in the network, the average number of neighbors is 13.416, the clustering coefficient is 0.457, and the network heterogenicity is 1.044. The top ten hub genes were TPX2, TTK, RRM2, TOP2A, NCAPG, CCNB2, ZWINT, CCNB1, CDK1, and RFC4. And after analyzing with The Cytoscape plugin Molecular Complex Detection (MCODE; version 1.6) of Cytoscape (3.7.2), we got three statically and medically significant modules. We analyzed all three modules and selected the top five hub genes in module 1. They are TOP2A, CCNB2, TTK, PRC1, and NUSAP1 (Fig. 13a,b).
Prognostic survival analysis
We analyzed the effect of the top ten hub genes on the prognosis of HCC patients (Fig. 14, a-j). CCNB2 is the only gene with a Log-rank P value more the 0.05, which indicates that CCNB2 has almost no correlation with overall survival outcome of HCC. CDK1 and CCNB1 have the smallest Log-rank P-value among these (0.00017 and 0.00015, respectively), which indicates that both of these two genes have a strong correlation with the prognosis of HCC.
Expression level of CDK1 in different stages of HCC
The expression of CDK1 was different in different stages of HCC, with the highest expression level in stage 3 and the lowest in stage 4. The discrete degree of expression level reached the highest in the second stage. (Fig. 15)
Phylogenic tree of CDK1 family analysis
The phylogenic of the CDK1 protein family was based on the maximum-likelihood among species. From this phylogenic tree, we can easily conclude that human CDK1 gene and ape-like CDK1 gene subtype 5 may be homologous genes. However, human CDK1 gene and the gene encoding human CDK1 protein A chain may be heterozygous genes (Fig. 16).