Owen Powell

Predictive Breeding for Precision Pulses

Globally, demand for plant-based protein is increasing with more than 100,000 tonnes of pulse-based protein required by 2030. Despite the increasing demand for pulse-based protein, expansion of pulse crop production is hindered in Australia due to low baseline yield and high variability across seasons.

This project aims to use artificial intelligence algorthims to deconvolute complex relationships between genotype, the environment and phenotype to supercharge the development of improved pulse varieties for the future. The ability of deep learning algorithms to identify these complex network relationships will be benchmarked against existing predictive breeding methods using both in silico and experimental datasets.

In collaboration with wider QAAFI, UQ ARC Centre for Excellence for Plant Success in Nature and Agriculture and JLU research teams, the successful candidate will develop experience and skills in the use of simulation (digital twin) software, data science, predictive methods (machine learning, deep learning) and gene discovery as part of a research pipeline to deliver impact through enabling prediction-based pulse improvement. While there could be the potential to complement the evaluation of crop growth model enhanced genomic prediction against other statistical algorithms and targeted experiments on traits contributing to yield and yield stability for chickpea and/or mungbean in the UQ Plant Futures Facility. The weighting of computer versus experimental activities can be weighted to suit the successful candidate.

The successful candidate will develop broad skills and experience in data collection, quality control, curation, reproducible research documentation and analyses. So, although the direct results will be related to agriculture, the research skills to be investigated and learned are transferable to genomics and data science more widely.