skyline: Skyline Plot Estimate of Effective Population Size

Description

skyline computes the generalized skyline plot estimate of effective population size from an estimated phylogeny. The demographic history is approximated by a step-function. The number of parameters of the skyline plot (i.e. its smoothness) is controlled by a parameter epsilon. find.skyline.epsilon searches for an optimal value of the epsilon parameter, i.e. the value that maximizes the AICc-corrected log-likelihood (logL.AICc).

Usage

skyline(x, ...)
skyline.phylo(x, ...)
skyline.coalescentIntervals(x, epsilon=0, ...)
skyline.collapsedIntervals(x, old.style=FALSE, ...)
find.skyline.epsilon(ci, GRID=1000, MINEPS=1e-6, ...)

Arguments

Either an ultrametric tree (i.e. an object of class "phylo"), or coalescent intervals (i.e. an object of class "coalescentIntervals"), or collapsed coalescent intervals (i.e. an object of class "collapsedI

epsilon

collapsing parameter that controls the amount of smoothing (allowed range: from 0 to ci$total.depth, default value: 0). This is the same parameter as in collapsed.intervals

old.style

Parameter to choose between two slightly different variants of the generalized skyline plot (Strimmer and Pybus, pers. comm.). The default value FALSE is recommended.

coalescent intervals (i.e. an object of class "coalescentIntervals")

GRID

Parameter for the grid search for epsilon in find.skyline.epsilon.

MINEPS

Parameter for the grid search for epsilon in find.skyline.epsilon.

...

Any of the above parameters.

Value

skyline returns an object of class "skyline" with the following entries:
timeA vector with the time at the end of each coalescent interval (i.e. the accumulated interval lengths from the beginning of the first interval to the end of an interval)
interval.lengthA vector with the length of each interval.
population.sizeA vector with the effective population size of each interval.
parameter.countNumber of free parameters in the skyline plot.
epsilonThe value of the underlying smoothing parameter.
logLLog-likelihood of skyline plot (see Strimmer and Pybus, 2001).
logL.AICcAICc corrected log-likelihood (see Strimmer and Pybus, 2001).
find.skyline.epsilon returns the value of the epsilon parameter that maximizes logL.AICc.

Details

skyline implements the generalized skyline plot introduced in Strimmer and Pybus (2001). For epsilon = 0 the generalized skyline plot degenerates to the classic skyline plot described in Pybus et al. (2000). The latter is in turn directly related to lineage-through-time plots (Nee et al., 1995).

References

Strimmer, K. and Pybus, O. G. (2001) Exploring the demographic history of DNA sequences using the generalized skyline plot. Molecular Biology and Evolution, 18, 2298--2305. Pybus, O. G, Rambaut, A. and Harvey, P. H. (2000) An integrated framework for the inference of viral population history from reconstructed genealogies. Genetics, 155, 1429--1437. Nee, S., Holmes, E. C., Rambaut, A. and Harvey, P. H. (1995) Inferring population history from molecular phylogenies. Philosophical Transactions of the Royal Society of London. Series B. Biological Sciences, 349, 25--31.

Examples

Run this code

library(ape)

# get tree
data("hivtree.newick") # example tree in NH format
tree.hiv <- read.tree(text = hivtree.newick) # load tree

# corresponding coalescent intervals
ci <- coalescent.intervals(tree.hiv) # from tree

# collapsed intervals
cl1 <- collapsed.intervals(ci,0)
cl2 <- collapsed.intervals(ci,0.0119)


#### classic skyline plot ####
sk1 <- skyline(cl1)        # from collapsed intervals 
sk1 <- skyline(ci)         # from coalescent intervals
sk1 <- skyline(tree.hiv)   # from tree
sk1

plot(skyline(tree.hiv))
skylineplot(tree.hiv) # shortcut

plot(sk1, show.years=TRUE, subst.rate=0.0023, present.year = 1997)

#### generalized skyline plot ####

sk2 <- skyline(cl2)              # from collapsed intervals
sk2 <- skyline(ci, 0.0119)       # from coalescent intervals
sk2 <- skyline(tree.hiv, 0.0119) # from tree
sk2

plot(sk2)


# classic and generalized skyline plot together in one plot
plot(sk1, show.years=TRUE, subst.rate=0.0023, present.year = 1997, col=c(grey(.8),1))
lines(sk2,  show.years=TRUE, subst.rate=0.0023, present.year = 1997)
legend(.15,500, c("classic", "generalized"), col=c(grey(.8),1),lty=1)


# find optimal epsilon parameter using AICc criterion
find.skyline.epsilon(ci)

sk3 <- skyline(ci, -1) # negative epsilon also triggers estimation of epsilon
sk3$epsilon

Run the code above in your browser using DataLab

Data engineering and BI courses are free!