Fig. 1

The framework of detecting and verifying prognostic biomarkers of BRCA from gene expression data by the RCPH methods. a Download the publically available RNA-seq data and then select DEGs. b Add the prior knowledge from KEGG, GO, MammaPrint, OSbrca, and scPrognosis to DEGs and integrate all of them with RegNetwork to obtain a connected network component, and extract the gene expression values accordingly. c Apply the RCPH models on the network-structured data to select the feature genes of BRCA. d Choose the optimal feature subsets via the assessment of C-index and P-value. e Identify the genes with non-zero regression coefficients as the potential BRCA biomarkers. f Establish the PRS model based on statistically significant genes from the Cox model in order to make response predictions for prognosis and treatment of BRCA. g Perform survival analysis using the PRS index to investigate its prognostic performance. h Explore the significance and differences of PRS index in normal and tumor tissues