Update to plotting method for "multiSimmap" density

To better serve a web app that's being developed in my lab, I just added a new feature to the plot method associated with the object produced by a call of the function density on the "multiSimmap" object class.

By default, density.multiSimmap computes an object of class "changesMap" which can then be graphed to show a posterior distribution of character state changes of different types from the stochastic mapping.

Here's a quick example of what I mean.

First, I'll generate a set of stochastic maps for a discrete character.

## load phytools
library(phytools)
## version should be >=1.0.6
packageVersion("phytools")

## [1] '1.0.6'

## read tree from file
eel.tree<-read.tree(file=
    "http://www.phytools.org/Rbook/8/elopomorph.tre")
eel.tree

## 
## Phylogenetic tree with 61 tips and 60 internal nodes.
## 
## Tip labels:
##   Moringua_edwardsi, Kaupichthys_nuchalis, Gorgasia_taiwanensis, Heteroconger_hassi, Venefica_proboscidea, Anguilla_rostrata, ...
## 
## Rooted; includes branch lengths.

## read data from file
eel.data<-read.csv(file=
    "http://www.phytools.org/Rbook/8/elopomorph.csv",
    stringsAsFactors=TRUE,row.names=1)
head(eel.data)

##                   feed_mode Max_TL_cm
## Albula_vulpes       suction       104
## Anguilla_anguilla   suction        50
## Anguilla_bicolor    suction       120
## Anguilla_japonica   suction       150
## Anguilla_rostrata   suction       152
## Ariosoma_anago      suction        60

## extract discrete character
feed_mode<-setNames(eel.data$feed_mode,rownames(eel.data))
## stochastic mapping
simmap<-make.simmap(eel.tree,feed_mode,nsim=500,
    model="ARD",pi="fitzjohn")

## make.simmap is sampling character histories conditioned on
## the transition matrix
## 
## Q =
##                bite     suction
## bite    -0.01561726  0.01561726
## suction  0.01737975 -0.01737975
## (estimated using likelihood);
## and (mean) root node prior probabilities
## pi =
##      bite   suction 
## 0.5148674 0.4851326

## Done.

simmap

## 500 phylogenetic trees with mapped discrete characters

Now I'm going to compute the density "changesMap" object.

density<-density(simmap)
density

## 
## Distribution of changes from stochastic mapping:
##  bite->suction       suction->bite
##  Min.   :4   Min.   :4
##  Median :18  Median :15
##  Mean   :17.05   Mean   :15.58
##  Max.   :30  Max.   :30
## 
## 95% HPD interval(bite->suction): [7, 24]
## 95% HPD interval(suction->bite): [6, 25]

Finally, let's plot it.

plot(density)

What the update permits is each of the transition types to be graphed separately.

Here's what I mean. I'll actually combine them into a single figure, just to make it look a bit nicer!

par(mfrow=c(2,1))
plot(density,transition="bite->suction")
plot(density,transition="suction->bite")

To see what happens for a character with more than two states, I'm going to simulate a three level character the same eel.tree "phylo" object that we've been using.

Q<-matrix(c(
    -1,1,1,
    0,-1,1,
    0,0,0),3,3,byrow=TRUE,
    dimnames=list(letters[1:3],letters[1:3]))/100
Q

##       a     b    c
## a -0.01  0.01 0.01
## b  0.00 -0.01 0.01
## c  0.00  0.00 0.00

x<-sim.Mk(eel.tree,Q,anc="a")
simmap<-make.simmap(eel.tree,x,model="ARD",nsim=200,
    pi="fitzjohn")

## make.simmap is sampling character histories conditioned on
## the transition matrix
## 
## Q =
##            a          b         c
## a -0.0110744  0.0110744 0.0000000
## b  0.0000000 -0.0566945 0.0566945
## c  0.0000000  0.0000000 0.0000000
## (estimated using likelihood);
## and (mean) root node prior probabilities
## pi =
## a b c 
## 1 0 0

## Done.

simmap

## 200 phylogenetic trees with mapped discrete characters

This is what density.changesMap presently does by default with this kind of data.

plot(density(simmap))

Let's now apply what we just learned to show only changes between states 'a' and 'b', in both and backward directions.

par(mfrow=c(2,1))
plot(density(simmap),transition="a->b")
plot(density(simmap),transition="b->a")

All right.