Fig. 3

Construction of RNA modification characteristic signature. a Unsupervised clustering of the RNA modification phenotype-related genes. The names of CRC cohorts were used as sample annotations. Red, high expression of phenotype-related genes; blue, low expression. b-c Differences in the WM_Score between RNA modification patterns (b) and gene clusters (c) in eight GEO-CRC cohorts. The Wilcoxon test was used to determine the statistical significance of the difference, and p < 0.05 was considered statistically significant. d-e Kaplan-Meier curves show overall survival in WM_Score-high (red) and -low (blue) in the GSE39582 (d) and TCGA-COAD/READ (e) cohorts. The grouping of CRC samples is shown at the bottom of the chart. p < 0.05 in the two-sided log-rank test was considered statistically significant. f-g Multivariate Cox regression model analysis, which included the factors of WM_Score, patient age, gender, TNM status, and patient outcomes in the GSE39582 (f) and TCGA-COAD/READ (g) cohorts. The length of the horizontal line represents the 95% confidence interval (CI) for each group. The vertical dotted line represents the hazard ratio (HR) of all patients shown by the forest plot. h The predictive value of WM_Score in patients among the GSE39582 cohort (AUC:0.64, 066, and 0.64; 3, 6, and 12-months overall survival)