Fig. 6: Construction of radiation dose estimation models in mice and humans.
From: RNA N6-methyladenosine modification-based biomarkers for absorbed ionizing radiation dose estimation
a Fitting curves of two-order polynomial regression models for absorbed doses prediction in adult mice gamma-ray TBI model using the m6A percentages of Ncoa4 mRNA, which were measured by MeRIP-qPCR assay using the Ncoa4 primer 1. Actual irradiation doses were represented by scattered dots. X-axis: the observed m6A percentages of Ncoa4 in PBMCs of mice; Y-axis: absorbed doses of irradiation. TBI, total-body irradiation; TPI, time post irradiation (days). Error bands represent 95% confidence interval (CI) estimated by the two-order polynomial regression model. b Fitting surface of a binary three-order polynomial regression model (integrated model) for absorbed doses prediction in adult mice gamma-ray TBI model taking both m6A percentage of mouse Ncoa4 mRNA and TPI (days) as input. The m6A percentages of Ncoa4 mRNA were measured by MeRIP-qPCR using the Ncoa4 primer 1. X-axis: the observed m6A levels of Ncoa4 in PBMCs of mice; Y-axis: the observed time post irradiation (days); Z-axis: absorbed doses of irradiation. c Receiver operating characteristic (ROC) analysis of the performance of the integrated model on absorbed irradiation dose prediction in adult mice gamma-ray TBI model with fixed TPI (days). AUC, area under curve. d ROC analysis of the performance of the integrated model on absorbed irradiation dose prediction without TPI constrains in adult mice gamma-ray TBI model using the dose cutoff varying from 0.2 to 8 Gy. Fitting curves of two-order polynomial regression models for absorbed doses prediction in cancer patients receiving radiotherapy with multiple doses of X-ray exposure using NCOA4 primer 2 (e) and primer 3 (f). Actual irradiation doses were represented by scattered dots. X-axis: the observed m6A levels of NCOA4 in PBMCs of patients; Y-axis: the absorbed doses of irradiation. Error bands represent 95% CI estimated by the two-order polynomial regression model. (g and h) ROC analysis of the performance of the models on absorbed irradiation dose prediction in cancer patients receiving radiotherapy using NCOA4 primer 2 (g) and primer 3 (h). Source data are provided.
