Create step-changes in statistical methods to underpin R&D in sustainable primary industries
Mixed Models and Experimental Design Lab
Our people are statistical scientists who collaborate with industries and research providors in primary industries to develop statistical methods which have a measurable impact on the reliability of information obtained from the conduct of designed experiments. This in turn increases the confidence of stakeholders in this information and technologies which facilitates more rapid and widespread adoption.
The major focus of our work centres on plant improvement where our methods and software tools enable breeders to meet the challenges of improving productivity at the farm-gate in the face of rapidly changing environment and market scope.
Industry Research Collaborations
MMaED has collaborated with 13 industry researchers.
Statistical Software Solutions
MMaED has produced and collaborated on 6 statistical software solutions: Pedicure, ODW, DWReml, dwrPlus, ASRextras, ASReml-R
MMaED has published 400+ scientific publications, which is a testament to our remarkable dedication and invaluable contributions to the of utilization of linear and generalised linear mixed models, with particular application to plant and animal breeding data.
Years of Experience
With over four decades of experience, MMaED is at the forefront of utilizing mixed models and experimental design. We excel in improving statistical methodologies in the realm of plant and animal breeding data analysis, facilitating sustainable advancements in genetics.
DWReml is an R package that fits the general linear mixed model and estimates variance components by Residual Maximum Likelihood using the Average Information algorithm. DWRreml exploits a publicly available open source supernodal sparse linear equation solver (MUMPS: MUltifrontal Massively Parallel sparse direct Solver) to efficiently solve the mixed model equations. The package offers a functional style user interface supporting several common variance models and allows partitioning of the (possibly correlated) experimental units into independent sections.
ODW is an R package to generate optimal categorical experimental designs under a general linear mixed model. Given an initial configuration and an objective set of effects, ODW iteratively interchanges rows (experimental units) of the corresponding column partition of the incidence matrix in search of a permutation that minimizes the A-value for the objective effects. The package offers a functional style user interface similar to DWReml and a concise set of variance models.
Director: Senior Professor Brian Cullis
Phone: +61 2 4221 5641
Location: Building 39C Room 266
I am honoured to lead such a passionate, innovative and dedicated team of statistical scientists. Each member of the team make a difference each and every day with their commitment to providing relevant and bespoke statistical solutions to solve problems for the primary industries R&D sector.
Associate Research Fellow
Associate Research Fellow
- Rijk Zwaan Zaadteelt en Zaadhandel B.V.
- Department of Jobs, Precinct and Regions – National Lentil Breeding Program
- Rice Breeding Australia
- Australasian Plant Phenomics Facility – University of Adelaide
- Centre for Crop Science, Queensland Alliance for Agriculture and Food Innovation, The University of Queensland