A phylogenetic tree model on tree $T$ has outcomes that are described by assigning each leaf of the tree any label from a particular set (typically the label set is the set of DNA bases, \{A,T,C,G\}). The probability of a certain assignment of labels depends on transition probabilities between each ordered pair of labels. These transition probabilities are the parameters of the model.
In a group based model, the label set is a group $G$ (typically $\mathbb{Z}/2$ or $(\mathbb{Z}/2)^2$), and the transition probability for a pair $(g,h)$ depends only on $h-g$. This reduces the number of parameters from $|G|^2$ to $|G|$. Depending on the model, further identifications of parameters are imposed.
An object of class Model stores the information about a group-based model required to compute phylogenetic invariants. This information includes the elements of the group, how those elements are partitioned, and a set of automorphisms of the group that preserve the partitions.
There are four built-in models: Cavender-Farris-Neyman or binary model (CFNmodel); Jukes-Cantor model (JCmodel); Kimura 2-parameter model (K2Pmodel); and Kimura 3-parameter model (K3Pmodel). Other models can be constructed with model.
|
|
|