Momocs, morphometrics using R

Momocs is an R package intended to ease and to popularize morphometrics, the quantitative analysis of form. It aims at gathering the most common approaches: traditional morphometrics, global descriptors, open outlines, closed outlines and configuration of landmarks into the same, open-source, environment. And using the same, simple and easy-to-learn, open-source, and extendable grammar both for the user and programmers. Momocs hinges on the core functions developed in the must-have book Morphometrics with R by Julien Claude (2008), wraps them into the same coding grammar, and extends them both graphically and statistically.

Use Momocs

Get the last version:

• First get devtools with install.packages("devtools")
• Then, download and load the very last version with:

devtools::install_github("vbonhomme/Momocs", build_vignettes=TRUE)
library(Momocs)

Note that:

• The current CRAN version (0.2) has been published in Journal of Statistical Software
• Since then, Momocs has been almost entirely rewritten and considerably enlarged.
• A CRAN version will be released when most of the features planned below (*) will be part of Momocs.
• Also, some tutorials are on their way: browseVignettes("Momocs"). Same thing, here

You are welcome to:

• propose ideas and report bugs
• offer your published data to the world
• ask for hotline and collaborate with me: [email protected]

Features

Data acquisition + Babel

• Outline extraction
• Landmark definition on outlines* (via StereoMorph)
• Open curves digitization with bezier curves* (via StereoMorph)
• Import from .nts, .tps, PAST, .txt, etc.
• Export to .txt, etc.

Outline analysis

• Elliptical Fourier analysis (efourier)
• Radius Variation Fourier analysis (rfourier)
• Tangent Angle Fourier analysis (tfourier)
• Calibration for all methods (calibrate_reconstructions, calibrate_deviations, calibrate_harmonicpower)

Open-outlines

• Natural (raw) polynomials (npoly)
• Orthogonal (Legendre) polynomials (opoly)
• Discrete Cosinus Transform (dfourier)
• Same calibration methods as above

Configuration of landmarks

• Full Generalized Procrustes Adjustment (fgProcrustes)
• Resistant Fit Procrustes Adjustments*
• Sliding semi-landmarks*

Traditional morphometrics and global shape descriptors

• Some built-in facilities for multivariate analysis*
• A long list of shape scalars (eg. eccentricity, rectilinearity, etc.)

Data handling

• filter, select, slice, mutate ala dplyr
• split according to a factor (eg. a 2D view)
• combine several 2D views

Multivariate analysis

• Mean shape (groupwise) calculations (mshapes)
• Principal component analysis (PCA)
• Multivariate analysis of variance (MANOVA + pairwise testing MANOVA_PW)
• Linear discriminant analysis (LDA)
• Hierarchical clustering (CLUST)
• Regression trees*
• K-means*
• Multidimensional scaling*
• Allometry / modularity testing*

Graphical methods

• Elegant ggplot2 plots
• Morphological spaces, when possible, for multivate analyses (and for combined views)
• Thin plate splines and variation around deformation grids
• Family pictures and quick inspection of whole datasets (panel, stack, plot, radar, etc.)

Various

• Toy and published datasets
• Vignettes
• Shiny demonstrators/helpers

Architecture

Here is a scheme of the Momocs' architecture: