Fig. 5.

Combining OAM and tumor genome has value in predicting patient prognosis. A C-index of “coxnet”, “random forest”, “xgboost” and “coxboost” in BRCA, UCEC, CESC, LUSC, LUAD, KIRC, GBM, HNSC, THCA, BLCA, STAD, COAD, ESCA, OV, READ, KIRP and KICH. B C-index (x-axis) of the best model for in LUSC, UCEC, GBM, READ, LUAD, BRCA, BLCA, COAD, HNSC, ESCA, CESC, STAD, OV, KIRP, KIRC, THCA and KICH. Different tumor types are distinguished by different colors. C AUROC values (x-axis) of time-ROC at 1, 3 and 5 years for the best model of 17 tumor types. Red, blue and green represent 1 year, 3 years and 5 years respectively. D CARS score of the highest prognosis-related microorganisms in KM curves of 17 tumor types. The subgroups were distinguished by the best cut-off value and the log-rank test was used to calculate the p value. The high relative abundance group is shown in yellow and the low relative abundance group is shown in blue. OAM oncogenome associated microbiome module, C-index Harrell’s Consistency Index, TCGA the cancer genome atlas, BRCA breast cancer, UCEC endometrioid cancer, CESC cervical cancer, LUSC lung squamous cell carcinoma, LUAD lung adenocarcinoma, KIRC kidney clear cell carcinoma, GBM glioblastoma, HNSC head and neck cancer, THCA thyroid cancer, BLCA bladder cancer, STAD stomach cancer, COAD colon cancer, ESCA esophageal cancer, OV ovarian cancer, READ rectal cancer, KIRP kidney papillary cell carcinoma, KICH kidney chromophobe, AUROC the area under the curve of the receiver operating characteristic, ROC the receiver operating characteristic, CARS correlation-adjusted regression survival, KM Kaplan–Meier