Extended Data Fig. 2. Additive model of driver advantages.

(a) Schematic figure of probabilistic growth of tumour voxels, with the growth probability of a tumour voxel cumulatively determined by the drivers harboured. (b) A table summarising the assumed amount of growth probabilities added by each driver ($p_{growth\_k}$ for a driver k). Three scenarios were explored, indicated in the table by $p_{growth\_k}$ of the weakest driver, namely, min($p_{growth\_k}$), and the difference in $p_{growth\_k}$ between consecutive two drivers as in the strength of their association with Ki67 IHC score, namely, $\mathrm{\Delta }{p}_{growth\_k}$. (c-d) Whole-tumour cancer cell fraction (CCF) of parental and largest subclones in in-silico tumours under Volume Growth (c) and Surface Growth (d), respectively, under the indicated parameter conditions. ‘Parental (3p loss, VHL)’ clone is shown along with up to five subclones with a whole-tumour CCF of 0.01 or higher. All remaining subclones are represented in the ‘other’ group. N = 100 for each condition.

Source data