Divergence time based on extrapolation

This method infers the divergence time for nodes lacking them based on the divergence times from other nodes in the tree.

Usage

inferDivTimesFromExtrapolation(
  pathToRoot,
  cladeName = NULL,
  verbose = FALSE,
  models = c("t_gamdec", "t_lin", "t_scam", "t_exp", "t_gam", "t_iso", "t_power")
)

Arguments

pathToRoot

A ConvenientTblTree object containing all nodes along the path from a tip to the root of the tree.

cladeName

An optional parameter that describes the dataset.

verbose

Boolean flag (optional; default value is FALSE). If TRUE, the function prints messages about the computation and summary statistics for the dataset.

models

An optional parameter defining which models will be used for extrapolating the data from TimeTree. Notice that only models with sufficient data are included in the results; the amount of information required varies by model. Available options: t_gam (Generalized Additive Model),t_scam (Shape Constrained Additive Model), t_gamdec (Generalized Additive Model with a decreasing monotonicity constraint), t_iso (Isotonic regression), t_lin (Linear interpolation), t_exp (Exponential model),t_power (Power-law model).

Value

A named list with the following: 1. models: a named list of the models actually used to extrapolate the data from TimeTree. A model passed via the input parameter models is omitted if there is insufficient data to use it. NULL is returned if no model was used. 2. path: the ConvenientTblTree object received as input with divergence-time estimates assigned to ancestral nodes that lacked them. 3. known: nodes used as the basis for inference. 4. unknown: nodes whose divergence times were inferred. 3. x_grid: Extra points spanning the region of known and unknown points where the models were evaluated. 4. prediction_grids: A named list with one element per model, containing predictions for the unknown points (unknown) and the extra points (x_grid).

Details

The method first finds a function describing the relationship between cumulative branch length and divergence time. It then uses this function to estimate divergence times for nodes that lack them, given their cumulative branch lengths.

Note

See functions model_gam (option t_gam), model_scam (option t_scam), model_gamdec (t_gamdec), model_iso (option t_iso), model_lin (option t_lin), model_exp (option t_exp), model_power (option t_power) for more details on each model.

As the relationship between cumulative branch length and divergence time is not necessarily linear and varies for each clade, the model that seemed to best fit different cases is the Generalized Additive Model (GAM) with a decreasing monotonicity constraint (t_gamdec).

Author

Priscila Biller