Generalized Additive Model (GAM) with a decreasing monotonicity constraint.

This function uses the relationship between cumulative branch length and time to infer the divergence times of ancestral nodes that have only the cumulative branch length.

Usage

model_gamdec(known, unknown, x_grid = NULL)

Arguments

known

Nodes with both divergence time (column divergenceTime) and cumulative branch length (column cumBranchLen).

unknown

Nodes with only cumulative branch length (column cumBranchLen; column divergenceTime is NA).

x_grid

Optional parameter with extra points beyond those with unknown divergence times. These can be used, for example, to plot or check accuracy over a wider range.

Value

A list with 2 types of predictions: 1. pred_unknown: predictions for nodes in the path with unknown divergence time; 2. pred_grid: predictions for the extra points passed in x_grid.

Details

The relationship between cumulative branch length and time is initially inferred with GAM (Generalized Additive Model), and later the constrain monotonic decrease is added.

Note

This function is similar to model_scam (same principle but different results).

Author

Priscila Biller