1. Data acquisition
RNA-sequencing expression profiles and the corresponding clinical information of ovarian cancer were downloaded from TCGA (http://portal.gdc.cancer.gov/). Patients without survival information were excluded. A total of 22 disulfidptosis-related genes were selected from previous studies5, including FLNA, FLNB, MYH9, TLN1, ACTB, MYL6, MYH10, CAPZB, DSTN, IQGAP1, ACTN4, PDLIM1, CD2AP, INF2, SLC7A11, SLC3A2, NUBPL, NDUFA11, LRPPRC, OXSM, NDUFS1, and GYS1.
2. Establishment of disulfidptosis-related cluster
We used the “ConsensusCluster Plus” package for consensus clustering analysis and identified clusters of ovarian cancer patients based on disulfidptosis-related genes. T-distributed Stochastic Neighbor Embedding (t-SNE) analysis was performed to evaluate the difference between two clusters. Additionally, Kaplan–Meier survival analysis evaluated the overall survival (OS) of different clusters, and the heat map showed the difference in disulfidptosis-related genes between clusters.
3. Time dependent receiver operating characteristic (ROC) analysis
Time dependent ROC curves were calculated using the timeROC (v.0.4) R package to predict the 1-year, 3-year, and 5-year outcomes of ovarian patients.
4. Identification of DEGs and functional enrichment analysis
The Limma package was used to identify DEGs between two clusters. “Adjusted P < 0.05 and |Fold Change|>1.5” were defined as the threshold for identifying the DEGs. Heatmaps and volcano plots were generated using the R packages pheatmap and ggplot2, respectively. Gene Ontology (GO) and pathway analysis were utilized to explore the biological functions of DEGs.
5. Construction of the risk model
Univariate Cox regression analysis was used to preliminarily screen survival-related genes (p < 0.05), and then multivariate Cox regression was used to obtain the final candidate genes. In accordance with the median risk index, ovarian cancer patients were divided into high/low -risk group for subsequent analysis. The difference in survival between the two groups was compared by Kaplan–Meier curve and log-rank test.
6. Gene mutation frequency analysis
Somatic mutation data of the ovarian cancer samples were obtained from TCGA GDC Data Portal in “maf” format. Waterfall plots were then performed using the “Maftools” package in R software, which facilitated the visualization and summarization of the mutated genes.
7. Analysis of tumor immune microenvironment
We used the CIBERSORT algorithm to calculate the proportion of tumor-infiltrating immune cells10. The difference in the proportion of tumor-infiltrating immune cells between the high- and low-risk groups was compared. The ESTIMATE algorithm was
used to evaluate the differences in immune, stromal, and ESTIMATE scores between the high- and low-risk groups11.
8. Prediction of drug sensitivity
The “oncopredict” R package was used to calculate the half-inhibitory concentration (IC50) value of the two groups for 198 drugs12. The differences in drug sensitivities between the high-risk and low-risk groups were compared using the Wilcoxon test.
9. Establishment of a nomogram
We use the "rms" package to develop a predictive nomogram. In the nomogram, each variable is matched to a score and the total score is obtained by summing all variables in each sample. The nomogram plot was evaluated using ROC curves.
10. Statistical analysis
We utilized R software (version 4.3.2) for all data analyses. Differences between two groups were analyzed using Student’s t-test or the Wilcoxon test. Furthermore, correlation analyses were conducted using Pearson correlation analysis. GraphPad Prism software (version 9) was used for plotting the images, with P < 0.05 as the threshold of significance for all statistical analyses.