Fig. 2. An example workflow of Butte.

Tumor ESCA_R_8 has a clonal SCNA (at 6:1) on chr13, as indicated on the top left. LogR is the Log2 copy number ratio between the tumor and matched normal sample and the B-Allele Frequency quantifies the allelic imbalances. Butte takes the read counts (total depth and the depth of the mutant allele) for SSNVs on chr13, the SCNA state and tumor purity as the input. Butte then works out the allele state distribution by using Expectation Maximization. By adopting linear programming with all possible history matrices, Butte returns the upper bounds of initiation and lead time of the SCNA, respectively (CI: confidence interval). The variant allele frequency distribution of SSNVs is shown on the bottom left to illuminate the relationship between allele states and SCNA timing. Note that the VAF is affected by tumor purity.