TTAT

Given a phylogenetic tree, we want to ask if more closely related taxa are more likely to share the same trait values than we would expect by chance. This R package provides analytical methods to decide if the association between tips and characters is significant. By traversing the whole tree, this package provides functions highlighting all the subclades in which tips and characters are significantly associated.

Installation

You can install the development version of TTAT from GitHub with:

# install.packages("devtools")
devtools::install_github("leke-lyu/TTAT")

Features

The Association Index (AI) By measuring the imbalance of internal phylogeny nodes, the AI statistic explicitly considers the shape of the phylogeny.

The Parsimony Score (PS) We use Sankoff’s Algorithm to calculate the parsimony score. Note that low PS scores represent robust phylogeny–trait association.

Traverse the Tree The recursive function travels all the internal nodes to deploy the test and to identify exciting subclades.

Example

Load TTAT package

library(TTAT)
## basic example code

Generate rtree sample and assign traits on the tips

library(ape)
library(magrittr)

tree <- rtree(50)
data <- data.frame(tip=tree$tip.label, num=sample(1:4, 50, replace = T))
data[,2] %<>% as.character()

plotTreeWithTrait(tree, data)

Generate the null distribution and deploy the test

traverseTest(tree, data, "AI", 2000) %>% testSummary()
#> 7 tests have been deployed on the input tree with 50 tips using AI
#> ----------------------------------------------
#>                 root
#> Statistic    4.52569
#> Significance 0.64950
#> ----------------------------------------------
#>                node60   node74   node73   node72   node59   node58
#> Statistic    1.492369 1.289120 1.539152 2.424243 3.916612 4.227327
#> Significance 0.695500 0.969500 0.986000 0.936500 0.977500 0.796500

Present the null distribution

traverseTest(tree, data, "PS", 2000) %>% plotNullModel()

Future steps

Parallel Processing To boost the speed when estimating the null distribution.

Extending the Choice of Metrics the phylogenetic diversity (PD), the net relatedness index (NRI), and nearest taxa index (NTI)