Multivariate Comparative Tools for Fitting Evolutionary Models to Morphometric Data

Fits multivariate (Brownian Motion, Early Burst, ACDC, Ornstein-Uhlenbeck and Shifts) models of continuous traits evolution on trees and time series. 'mvMORPH' also proposes high-dimensional multivariate comparative tools (linear models using Generalized Least Squares and multivariate tests) based on penalized likelihood. See Clavel et al. (2015) , Clavel et al. (2019) , and Clavel & Morlon (2020) .

mvMORPH: an R package for fitting multivariate evolutionary models to morphometric data

This package allows the fitting of multivariate evolutionary models (Ornstein-Uhlenbeck, Brownian motion, Early burst, Shift models) on species trees and time series. It also provides functions to compute log-likelihood of users specified models with fast methods (e.g., for Bayesian approaches or customized comparative methods), simulates correlated traits under various models, constrain various parts of multivariate models...

The package implement now efficient methods for high-dimensional multivariate comparative methods (mvgls) based on Penalized likelihood

The package is designed to handle ultrametric and non-ultrametric trees (i.e. with fossil species) and missing data in multivariate datasets (NA values), SIMMAP mapping of discrete traits, measurement error, etc...

See the packages vignettes for details and examples: browseVignettes("mvMORPH").

mvMORPH 1.1.0

This is the version 1.1.0:

Penalized Likelihood methods for High-Dimensional datasets (mvgls)
Multivariate GLS (mvgls)
PCA on GLS covariance estimate (mvgls)
Some minor bugs corrections (see NEWS)

TODO:

Incorporation of a tests-suite
Implement the sampler (upcomming mvMORPH)
Code improvements
Extend the shift model to TS
Improved mvOU model
Threshold model for categorical data

The current stable version of the mvMORPH package (1.0.9) is on the CRAN repository. https://cran.r-project.org/package=mvMORPH

Package Installation

You can install the package directly from gitHub through devtools:

library(devtools)

install_github("JClavel/mvMORPH", build_vignettes = TRUE)

(The installation may crash if your dependencies are not up to date. Note that you may also need to install Rtools to compile the C codes included in the package. For Windows and for Mac (and Tools )

Report an issue

Any bugs encountered when using the package can be reported here

Package citation

Clavel, J., Escarguel, G., Merceron, G. 2015. mvMORPH: an R package for fitting multivariate evolutionary models to morphometric data. Methods in Ecology and Evolution, 6(11):1311-1319. DOI: 10.1111/2041-210X.12420

Download version with appended supplementary material.

News

mvMORPH 1.1.0

+ mvgls: a multivariate GLS function based on ML (n<p) and PL (p>n).
+ mvgls.pca: a function to perform PCA on serially correlated data.
+ GIC: function to compute the GIC score for both ML and PL techniques (class mvgls)
+ suite of functions to use with mvgls
+ Bug fix for "mvSIM" with sparse methods and speed-up for "BM1" model.
+ Typo fix in testLRT
+ Typo fix in weight-matrix-mvmorph.c (indexing of variable)
+ add options for the pruning algorithm.
+ fix error with NA values for the mvOU function.

mvMORPH 1.0.9

+ Registration of compiled codes
+ New pruning algorithm to compute the determinant and square root matrix
+ Bugs fixes in "estim" for ancestral state reconstructions with OU models and "smean=FALSE" option.
+ Bug fix for "estim" with univariate data.
+ Corrected bug in mvSIM.
+ Fix bug with matrix parametrization "equaldiagonal".
+ add the "svd" option for simulating traits in mvSIM.
+ Bug fix in "mvEB".

mvMORPH 1.0.8

+ Allows estimating the missing cases (NA)
+ Allows estimating trends
+ User defined constrained models and parameterizations  
+ Return the log-likelihood function
+ Simulating traits on package vignette
+ Multivariate models for time-series (TS)
+ Partial implementation of a tests-suite