Fig. 1

Overview of the clonal tree inference process. From left to right: multiple samples are taken from a heterogeneous tumor, either from different anatomical sites or different times; the samples are sequenced, the resulting reads are aligned to a reference genome, and variants are called; the VAF matrix is built from the reference and variant read counts; we build an ancestry graph from the VAF matrix; each ancestry graph spanning tree that adheres to the sum condition is a candidate clonal tree, two of which are shown. Notice that the second tree could be discounted if we were aware of mutation co-occurrence, because the dark blue and green mutations always appear together in the tumor