/correlate

R package for character correlation analyses with phylogenetic trees

Primary LanguageR

correlate

correlate performes character correlation analyses of categorical (discrete) characters based on stochastic character mappings.

Change History

August 28, 2019

- added basic documentation for all functions in correlate.

July 12, 2017

- added some basic documentation for the three core functions get_intersect(), get_marginal() and get_conditional().

July 19, 2016

 - new version 0.2
 - added function to perform Pagel'S 1994 correlation test on multiple trees
 - functions for correlation test have been renamed for consistency
 - small bug fixes

April 17, 2016:

- added documentation for some functions

April 15, 2016:

- added correct behavior for single trees
- added function summarize_correlate() which produces basic summary statistics.
- added possibility to label axis in plot_correlation()
- some small code clean-ups

DESCRIPTION

This small package performs character correlation analyses based on stochastic character maps from two sets of categorical characters. This also works for sets of trees (e.g. from a Bayesian posterior tree sample), to account for phylogenetic uncertainty. The functions in correlate will calculate the conditional probability for character pairs occurring on a particular point on the phylogenetic tree and output them as summarized probability distributions in the analyzed tree space. This method refers to concepts described in this paper by Huelsenbeck et al. (2003)

REQUIREMENTS

  • R 3.2.2
  • R dependecies: devtools, phytools and reshape, ggplot2 (if you want to use the included plotting function)
  • one or more phylogenetic trees
  • characters to map

INSTALL

First you need to install and load the devtools R package:

install.packages("devtools")

library(devtools)

Now you can install correlate:

install_github("reslp/correlate")

USAGE

I am working on a detailed documentation of the functions in correlate. Meanwhile here is a very brief describtions of what steps are needed for correlate to run:

  1. Lets assume you have a set of 10 stochastic maps for each of 10 trees created with phytools for two character sets named map1and map2
  2. First load correlate: library(correlate)
  3. Now you can call the function correlate: corr_matrix <- get_intersect(ntrees=10, nmaps=10, chars2=c("bark","lichenicolous", "rock", "soil", "wood"), chars1=c(0,1),smap1=map1,smap2=map2) ntrees and nmaps specify the number of trees and stochastic maps, chars1 and chars2 which characters should be analyzed in a pairwise matter. smap1 and smap2 specify the stoachstic maps.
  4. To calculate the conditional you will also need the probability for each character to appear alone on the tree(s): prob <- get_marginal(simmap_multi). This is only needed for one of the mappings, depending on the question you want to ask. See the wiki on conditional probaility for further information.
  5. Now you can calculate the conditional probability: cond <- get_conditional(matrix=corr_matrix, probs=prob)
  6. correlate also provides the output of the obtained conditional probability distributions as Violin plots: plot_correlation(cond, title="title") (https://github.com/reslp/correlate/blob/master/correlate_example.png). This shows the conditional for binary characters 0 and 1 with the multistate character. For example the very left violin plot 0_bark could be read as: The probability of bark given that character 0 is observed.
  7. By running the built-in function summarize_correlate(cond) you will get basic information about the distributions including, means, variance and standard deviation. It will also compare distributions with pairwise wilcox tests.

Pagel's (1994) test:

This test can be performed with the function pagel_multi. It takes a list of trees and two character distributions for the tree tips as named vectors as input: pagel_multi(trees, character_a, character_b). It will fit two models to each tree in the list and perform chi-squared tests on log-likelihood differences of the models. The function returns an object of class "pagel", which includes log-likelihoods for both models, likelihood ratios and p-values of the chi-squared test. This object may be plotted with plot(pagel). Plotting requires ggplot2.

LIMITATIONS

Although functional, correlate is still in a very early development stage, so be careful when using this experimental software. I give no warranty! Also, there are several features I would like to include. Among those are:

  • Correlation of multiple (more then two) sets of characters
  • Improved plotting
  • higher speed and memory efficiency

COPYRIGTH AND LICENSE

Copyright (C) 2016-2019 Philipp Resl

This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

You should have received a copy of the GNU General Public License along with this program in the file LICENSE. If not, see http://www.gnu.org/licenses/.