FIG. 2.

Workflow for the simulation and validation. We separate the whole process into five main steps: in step 1, we randomly chose k total single cells from each region (indicated by the black frames), where we can pick $\widehat{k}$ dominant clones (indicated by red circles, also called true cells); in step 2, we simulated $\widehat{n}$ tumor samples from each region using the k cells; in step 3, we combined the $\widehat{n}$ tumor samples to get a simulated bulk tumor; in step 4, we deconvolved the bulk tumor integrating observed cells to get $k\prime =3\widehat{k}$ inferred clones; and in step 5, we assessed the performance using the $k\prime$ inferred clones and $3\widehat{k}$ true cells.