3.1 Identification of differentially expressed genes and classification of patients
To investigate the expression of genes associated with copper death in HCC, we compared relevant genes in normal and tumour tissues. The results revealed that LIAS, LIPT1, DLD, DLAT, PDHA1, PDHB, MTF1, GLS, and CDKN2A were highly expressed in tumour tissues and FDX1 in normal tissues (figure 2a, 2b). We also showed the correlation between the 10 genes (figure 2c). Based on the 10 genes, we divided the patients into three categories (figure 2d) and analysed the overall survival (OS) between them and found P<0.001 (figure 2e).
3.2 Development and validation of a prognostic model for copper death and SARS-CoV-2 association
The analysis identified 2443 differentially expressed genes between the three patient groups, which were found to be mainly associated with T-stage, stage, and gender (figure 3a). These genes were intersected with SARS-CoV-2-related genes, and there were 275 intersected genes in total (figure 3b). A single gene prognostic regression analysis was done based on the 275 genes and all 62 genes P<0.01 were associated with prognosis (figure 3c). Using these 62 genes for lasso regression analysis and lasso regression models (figure 3d, 3e), four genes were finally identified as our prognostic model risk score = SSX2IP*0.00100336665126057+RNF145*0.040899709061588+GTSE1*0.120514296049806+SLC1A5*0.166208688574242. Patients with TCGA and GEO were scored according to these four genes. The difference in their OS was by significance, P<0.001 for the training group (figure 3f),P=0.01 for the validation group (figure 3g).
3.3 Analysing the effects of prognostic models
The training and validation sets were analysed for area under the curve (AUC) and the AUC for the training set was found to be 0.804 (figure 4a), a very high value, and 0.676 (figure 4b) for the validation set. PCA (figure 4c-d) and t-SNE (figure 4e-f) analyses of patients in the high-risk and low-risk groups demonstrated a significantly different distribution of patients in the two groups. The validity of our model was further confirmed. Finally, we did a single-factor regression analysis (figure 4g) and a multi-factor regression analysis (figure 4h) of the riskscore. Again, it was validated that our riskscore can be used as an independent predictor of patient prognosis as well as clinical characteristics.
3.4 Clinical correlation and associated pathway analysis of 4 genes
SSX2IP, RNF145, GTSE1, SLC1A5 major T staging, stage staging, grade grading, gender related (figure 5a) were found. After this, all differential genes were analysed for related pathways and found to be enriched in the Biological Process mainly in the fatty acid metabolic, mitotic nuclear division pathway, and in the Cellular Component mainly in the chromosome, The cellular component was mainly enriched in chromosome, centromeric region, collagen-containing extracellulae. oxidoreductase activity, acting on CH-OH group of donors was enriched in Molecular Function (figure 5b, 5c). In KEGG-related pathways, the main focus is on Metabolism of xenobiotics by cytochrome P450, Drug metabolism - cytochrome P450, etc. (figure 5d, 5e ).
3.5 Analysis of immune infiltration of TCGA and GEO
Analysis of 16 immune cells and corresponding immune pathways in patients with TCGA showed that aDCs, iDCs, macrophages, neutrophils, NK cells, T helper cells, Tfh, Th2 cells, TIL and Treg cells were significantly elevated in the high-risk group (figure 6a). . There were also corresponding differences in 11 of the 13 immune pathways (figure 6c). This suggests that immune factors may be an important factor in the prognosis of patients. Correspondingly, five immune cells were also upregulated in the high-risk group in patients with GEO (figure 6b), and eight immune-related pathways were also differentially represented (figure 6d).
3.6 Predicting drug sensitivity and drug correlation analysis
By predicting 12 drugs, we found that six were more effective for the low-risk group (figure 7a-f) and, similarly, six were more effective for the high-risk group (figure 7g-l). Drug correlations were analysed for four genes and figure 8 showed that 16 drugs were associated with different genes.