BackgroundIncreasing evidence has indicated that abnormal epigenetic factors such as RNA m6A modification, histone modification, DNA methylation, RNA binding proteins and transcription factors are correlated with hepatocarcinogenesis. However, it is unknown how epigenetic modification-associated genes contribute to the occurrence and clinical outcome of hepatocellular carcinoma (HCC). Thus, we constructed the epigenetic modification-associated models that may enhance the diagnosis and prognosis of HCC.MethodsIn this study, we focused on the clinical value of epigenetic modification-associated genes for HCC. Our gene expression data were collected from TCGA and HCC data sets from the GEO database to ensure the reliability of the data. Their functions were analyzed by bioinformatics methods. We used lasso regression, Support vector machine (SVM), logistic regression and Cox regression to construct the diagnostic and prognostic models. We also constructed a nomogram of the practicability of the above-mentioned prognostic model. The above results were verified in an independent liver cancer data set from the ICGC database and clinical samples. Furthermore, we carried out pan-cancer analysis to verify the specificity of the above model and screened a wide range of drug candidates.ResultsMany epigenetic modification-associated genes were significantly different in HCC and normal liver tissues. The gene signatures showed a good ability to predict the occurrence and survival of HCC patients, as verified by DCA and ROC curve analysis.ConclusionGene signatures based on epigenetic modification-associated genes can be used to identify the occurrence and prognosis of liver cancer.

No abstract

Background Heat shock proteins (HSPs) are widely involved in tumor occurrence and development and are prognostic markers for multiple tumors. However, the role of HSPs in clear cell renal cell carcinoma (ccRCC) remains unclear. Methods We used Cytoscape to identify hub genes in the ccRCC single-cell sequencing data set from the Gene Expression Omnibus (GEO) data repository. We identified subtypes, C1 and C2, of The Cancer Genome Atlas (TCGA) patients based on the expression of hub genes using unsupervised consensus clustering. Principal component analysis (PCA) was used to verify the clustering differences, and Kaplan–Meier (K-M) estimate was used to verify the survival differences between C1 and C2 patients. We used TIMER 2.0 and CIBERSORT to evaluate the immune cell infiltration of HSP genes and C1 and C2 patients. The R package “pRRophetic” was used to evaluate the sensitivity in C1 and C2 patients to the four first-line treatment drugs. Results We identified six hub genes (HSP90AA1, HSPH1, HSPA1B, HSPA8, and HSPA1A) encoding HSP, five of which were significantly downregulated in TCGA group, and four had a protective effect on prognosis (p <0.05). Survival analysis showed that C1 patients had a better overall survival (p <0.001). TIMER 2.0 analysis showed that three HSP genes were significantly correlated with the infiltration of CD4+ T cells and CD4+ Th1 cells (|cor|>0.5, p<0.001). CIBERSORT showed significant differences in multiple infiltrating immune cells between C1 and C2 patients. Meanwhile, the expression of PD1 was significantly lower in C1 patients than in C2 patients, and the expression of PDL1 is the another way around. Drug sensitivity analysis showed that C1 patients were more sensitive to sorafenib, pazopanib, and axitinib (p <0.001). Conclusion Our research revealed two molecular subtypes of ccRCC based on 6 HSP genes, and revealed significant differences between the two subtypes in terms of clinical prognosis, immune infiltration, and drug sensitivity.