This is a project which is currently making use of HPC facilities at Newcastle University. It is active.
For further information about this project, please contact:
This project investigates whether morphological character data used in palaeontology and systematics contain systematic patterns of character co-variation that are not adequately represented by conventional phylogenetic models. The project uses mutual information (MI) between morphological characters to construct character-distance spaces and identify structural patterns within empirical datasets.
The research will compare empirical morphological datasets with data simulated under the Mk model and related models. Analyses will use multidimensional scaling (MDS), distance-distribution statistics, minimum spanning trees and other geometric/skeleton-based methods to characterise the structure of character spaces. A major component is posterior-predictive model checking, assessing whether candidate evolutionary models can reproduce the structural properties observed in real morphological datasets.
The project will primarily use R and the AutoPart research software package. The computational workflow includes calculation of pairwise Adjusted Mutual Information (AMI) and Normalised Variation of Information (NVI) distances between morphological characters, multidimensional scaling, intrinsic-dimensionality analysis, and geometric/skeleton-based analyses including minimum spanning trees and related backbone methods.
The HPC facility will be used for computationally intensive analyses that are unsuitable for local execution, particularly posterior-predictive simulation/analysis and Bayesian MCMC production runs. The project includes Mk simulations and partitioned Bayesian analyses using the MkPrime package, with multiple treatments and large numbers of posterior trees.
The workflow will involve running multiple independent analyses across morphological datasets and comparing their resulting summary statistics and posterior distributions.