Background
The aim of this study was to utilize machine learning techniques to identify biomarkers associated with the diagnosis of bladder cancer, providing valuable insights into its early pathogenesis and exploring their potential as prognostic markers and therapeutic targets.
Methods
Initially, we conducted a comparative analysis of the genomes between bladder cancer samples, focusing on identifying the most significant differences between the cancer group and the normal group. Next, we employed machine learning techniques for feature selection and identified a key gene by integrating ferroptosis-related genes into our analysis. Moreover, we integrated transcriptome data, somatic mutation data, and clinical data to perform comprehensive analyses, including functional enrichment analysis, tumor mutation load analysis, immune infiltration analysis, and pan-cancer analysis. These analyses aimed to elucidate the pathological relevance of the candidate genes. Furthermore, we constructed a ceRNA network to identify the genes and regulatory pathways associated with these candidate genes.
Results
We initially conducted screening using the Weighted Gene Co-expression Network Analysis and machine learning techniques, resulting in the identification of six candidate genes: NR4A1, PAMR1, CFD, RAI2, ALG3, and HAAO. Subsequently, by integrating data from the FerrDB database, we identified NR4A1 as a gene associated with ferroptosis. Additionally, our analysis revealed a correlation between the expression of NR4A1 and tumor mutations as well as immune infiltration in patients with bladder cancer.
Conclusion
Our data strongly suggest that NR4A1 could serve as a crucial prognostic biomarker for bladder cancer and may also play a role in the development of various other cancers.