Fig. 4.

The results of sensitivity analysis assuming that tumours may originate from a hierarchy of cells. a bar chart of the estimated $R_i^2$ from the curve $g(LC{R}_i)={\gamma }_{0i}+{\gamma }_{1i}LTC{D}_0+{\epsilon }_{2i}$ in each EH_{lat i} ($i=1,\cdots ,423$) using global-wide ranked LCR matrix, representing the contributions of LTCD₀ from the optimal EH_lat (left) to the worst EH_lat (right) to the variation of cancer risks; b bar chart of the estimated $R_i^2$ from curve $g(LC{R}_i)={\lambda }_{0i}+{\lambda }_{1i}SM{N}_0+{\epsilon }_{3i}$ in each EH_{lat i} ($i=1,\cdots ,423$) using global-wide ranked LCR matrix, representing the contributions of SMN₀ from the optimal EH_lat (left) to the worst EH_lat (right) to the variation of cancer risks. LTCD₀: the error-prone value of the total number of tissue cell divisions per lifetime calculated in the laboratory environment; SMN₀: the error-prone value of the somatic mutation number calculated in the laboratory environment; LCR: the observed lifetime cancer risk, EH_lat: the level of EH (row) in the ranked LCR matrix