Fig. 7

Construction of the convenient SIPS prognostic model in ovarian cancer. A The Venn diagram showed the common genes of the four machine-learning methods (Boruta, Xgboost, Random Forest, and Lasso). B The boxplot showed the levels of SIGPS in patient with different MFP subtypes in TCGA-OV dataset. C The network plot showed the enriched functional modules of SIGPS genes. D The UMAP plot showed the levels of the SIGPS in tumor microenvironment in GSE165897 dataset. The SIGPS level of each cell was calculated by the AddModuleScore function based on R package Seurat. E The decision tree of HRD and SIGPS in TCGA-OV cohort. In the tree, Node 3: HRD⁺SIGPS⁻; Node 4: HRD⁺SIGPS⁺; Node 6: HRD⁻SIGPS⁻; Node 7: HRD⁻SIGPS⁺. F The boxplot showed the estimate score levels of Estimate Score, Immune Score, and Stroma Score among the four subgroups of the decision tree in TCGA-OV dataset. G The boxplot demonstrated the normalized camp scores of agents suppressing the SIGPS, and the horizontal axis showed the agent types. Each dot in the plot represented one agent. H The heatmap showing the expression levels of stroma-related genes, including TGFBI, FBLN2, and COL16A1, in the CAFs after the treatment of several drugs as indicated