Fig. 2

Aggregating the relative timing of events across samples. Once the timing of events within single samples has been established, partial orderings may be aggregated across a cohort to determine an average sequence of events. a Example phylogenetic trees which may be reconstructed from tumor life history analyses. Mutations A–D are highlighted on the tree based on their clonal frequency. b The outcome of pairwise comparisons between events within each sample, with the “winner” being the earliest event. Comparisons marked “NA” indicate cases where both events are present, but cannot be timed relative to one another. The final ordering: A and C cannot be timed against one another, but win against every other event, indicating that they are the earliest. B has an intermediate ranking, often earlier than D, but never before C or A. D is estimated to be last, as it only once wins a comparison (with B)