v1.8 release

[TGuillerme]

dispRity v1.8 (2023-12-11) dispRity.multi

NEW FEATURES

• Added the dispRity.multi internal architecture now allowing users to work with different matrices and different trees as inputs for custom.subsets, chrono.subsets, boot.matrix and dispRity. This change is not affecting the user level appart from now allowing to bypass some error messages (thanks to Mario Corio for that obvious suggestion).
• New utility function: name.subsets for directly accessing the subsets names of a dispRity object (basically doing names(my_dispRity$subsets)).
• New utility function: MCMCglmm.variance for calculating the variance for specific terms in a "MCMCglmm" model.
• New statistical test: pgls.dispRity to run PGLS test on a dispRity object with a level-2 metric and a tree (using excellent phylolm algorithm). The new test comes with its own S3 print, summary and plot functions if the input dispRity data contains multiple trees or multiple matrices (running and handling the output of multiple phylolm).
• New options to get.tree utility function to get the trees in each subsets (thanks to Jack Hadfield for this suggestion and help with implementation).
• New vignette compiling resources for developers to help people (and future me) to edit the package.
• New metric: roundness to measure the roundness of a variance-covariance matrix.

MINOR IMPROVEMENTS

• CHANGE IN DEFAULT ARGUMENTS for the custom.subsets, chrono.subsets, boot.matrix and dispRity functions: the tree, FADLAD, prob and dimensions arguments is now NULL by default (instead of missing).
• CHANGE IN DEFAULT ARGUMENTS for the projections function. The defaults are now scale = TRUE, centre = FALSE (previously TRUE) and abs = FALSE (previously TRUE). The default arguments for dispRity.covar.projections remain unchanged though (scale = TRUE, centre = TRUE, abs = TRUE).
• randtest.dispRity function's subsets argument now can take a list of arguments of subsets to compare is data is a dispRity object. The call message has also been updated to be much more verbose and clear of what has been tested.
• optimised internal logic for all the projections.* functions for speed.
• test.metric now also adds the slope coefficient if the option model is used.
• the default legends/row names when using plot/summary on test.metric results are now more easy to interpret.
• reduce.space has now 25 times less chances to be stuck on a local optimum.
• match.tip.edges now also works on nodes and on "multiPhylo" objects and has a added an option use.parsimony (default is TRUE) to propagate the matching down the tree (i.e. if two nodes have the same unique ancestor and variable value, the ancestor is consider to have that variable value as well).
• Improved margins handling in covar.plot.
• Updated test coverage here and there.
• clean.data can now also handle a list of "phylo" objects as tree input (i.e. it does not need to be specifically classed as "multiPhylo").
• Changed dependencies to spptest to GET.
• space.maker can now generate specific row names and be replicated to generate a bunch of spaces (via the new optional arguments elements.names and replicates respectively).
• The add.tree utility function now has an optional argument to override any existing trees (replace = TRUE) or not (replace = FALSE; which remains the default previous behaviour).
• print.dispRity now recognises and highlights simulated data from the treats package.
• Improved NA management for S3 functions.
• Improved automatic centering and scaling for covar.plot making the figures more aesthetic.
• remove.zero.brlen now also removes negative branch lengths and works on "multiPhylo" objects (thanks to Thomas Johnson for this suggestion).
• fill.dispRity has now an extra argument check to toggle the data checking on and off (more for developers).
• multi.ace is now described in details in the manual.
• slice.tree has now a keep.all.ancestors option to do exactly that.

BUG FIXES

• The correct types of changes are now plotted in legend when plotting the results of test.metric.
• get.disparity now correctly concatenates one dimensional results into a "numeric" (rather than a "matrix").
• make.metric now internally handles covar object correctly (i.e. as distance matrices).
• Calculating disparity for multiple matrices and customised subsets now works as expected for all metric levels.

DEPRECATED AND DEFUNCT

• ellipse.volume has been changed to ellipsoid.volume to more accurately reflect what it is measuring.
• rescale.dispRity has been changed to scale.dispRity and correctly registered as a S3 method.
• randtest.dist has been changed to distance.randtest to avoid conflict with *.dist S3 methods.