TO-DAG is an inferential method that deduces from cross-sectional data of genetic alterations in tumor patients the causal dependencies and the waiting times among these genetic events. From matrices with genetic events and patient samples as rows and columns, respectively, TO-DAG generates a probabilistic graph model whose nodes represent genetic events and oriented edges between nodes indicate the presence and the direction of a causal dependency between the nodes. A direct acyclic graph, i.e. a graph with no directed cycles, has been specifically chosen as model of putative causal dependencies, as genetic alterations are assumed to be irreversible events. Two parameters define an edge: (i) its probability estimated from the frequency of occurrence of the genetic events represented by the nodes and its conditional probability, and (ii) the waiting time, i.e. the time elapsing from the occurrence of a mutation to the occurrence of another one that is conditionally dependent on it.