Spain loans data science talent to Australian grains industry

Master of Biostatistics Program builds R&D capacity in computation and statistics

Learning & Development
University of Adelaide Biometry Hub leader Dr Olena Kravchuk says the Master of Biostatistics Program has helped build computation and statistical capacity for the Australian grains industry. PHOTO Clarisa Collis

University of Adelaide Biometry Hub leader Dr Olena Kravchuk says the Master of Biostatistics Program has helped build computation and statistical capacity for the Australian grains industry. PHOTO Clarisa Collis

Aa

Scholar program transports specialised skills and knowledge from Valencia to Adelaide.

Aa

True to its original Latin name Valentia meaning 'strength', the Spanish metropolis of Valencia is helping to reinforce the R&D capacity of the Australian grains industry.

There, on the banks of the Turia River beside the Mediterranean Sea, one of the oldest universities in Spain - the University of Valencia - is loaning its data science scholars to this country's grains research community.

Enabling the transfer of statistical science talent from the University of Valencia to the University of Adelaide, South Australia, is GRDC investment in a Master of Biostatistics Program.

Under the program, the University of Adelaide's dedicated centre for statistical and computational methods used in the design and analysis of biological experiments - the Biometry Hub - hosted University of Valencia master's-degree student Iker Oyanguren Monferrer for eight weeks from March to May 2019.

Biometry Hub leader Olena Kravchuk says the work placement, equivalent to a research assistant position, has already delivered better efficiency and consistency in approaches to computation and statistics research.

These improvements have contributed to an overall lift in plant pre-breeding selection and methodology, Dr Kravchuk says.

University of Valencia master's student Iker Monferrer used Bayesian logic to develop software able to better predict the influence of genetic and environmental influences in wheat pre-breeding as part of his work placement. PHOTO Sharon Watt

University of Valencia master's student Iker Monferrer used Bayesian logic to develop software able to better predict the influence of genetic and environmental influences in wheat pre-breeding as part of his work placement. PHOTO Sharon Watt

For instance, Iker was able to identify and overcome a bottleneck in the computational method used to check data quality, which means this preliminary step in statistical software development now takes three hours instead of three days.

His work designing new software has also helped improve the "creation, simulation and implementation of statistics" as part of wheat pre-breeding research, Dr Kravchuk says.

To achieve this, Iker created a software program applying a particular mathematical model - Bayesian statistics - to data sets sourced from a wheat pre-breeding program led by University of Adelaide Professor Diane Mather.

Bayesian probabilities

Dr Kravchuk says Iker's eight-week work placement at the Biometry Hub has contributed to an overall lift in wheat pre-breeding research selection and methodology. PHOTO Clarisa Collis

Dr Kravchuk says Iker's eight-week work placement at the Biometry Hub has contributed to an overall lift in wheat pre-breeding research selection and methodology. PHOTO Clarisa Collis

The Bayesian model, in which probabilities are applied to statistical problems, and conclusions are subjective and updated as additional data is collected, was used to determine the relative influence of a range of variables, both genetic and environmental, in the development of new wheat cultivars.

For example, working as a Bayesian data scientist at the Biometry Hub, Iker first analysed a sample data set from Professor Mather's wheat pre-breeding program to inform subjective conclusions, known as 'prior inference'.

He then analysed another sample data set from the same pre-breeding program to revise his conclusions, known as 'posterior inference'.

Drawing on this Bayesian logic, Iker generated software able to better predict, and ultimately rank, the influence of different variables, such as genetic parentage, agronomic practices and microclimate, within the University of Adelaide pre-breeding program.

Dr Kravchuk says applying the new software to plant pre-breeding projects more generally has the potential to improve the quality of the data science underpinning the scientific rigour of a range of research investments.

Program popularity 

Researchers from the SAGI South team based at the University of Adelaide's Biometry Hub: Olena Kravchuk, left, Dr Julian Taylor and Iker Oyanguren Monferrer. PHOTO Sharon Watt

Researchers from the SAGI South team based at the University of Adelaide's Biometry Hub: Olena Kravchuk, left, Dr Julian Taylor and Iker Oyanguren Monferrer. PHOTO Sharon Watt

Dr Kravchuk says that following Iker's research tenure, another three University of Valencia master's-degree students, including a mathematician, informatician and biostatistician, have applied for future work placement at the Biometry Hub.

The expertise of these students could be used to further develop Iker's work and explore computational and statistical efficiency gains in other areas of grains research, such as agronomy and pest control, she says.

"The work placement counts towards the training and qualification of a master's-degree student and helps to create a global pool of statisticians with skills and knowledge specialised to the grains industry and the wider agricultural sector," she says.

Also supporting the Master of Biostatistics Program was University of Valencia Department of Statistics and Operations Research leader Anabel Forte, who trained Iker in the software used at the Biometry Hub in an effort to optimise the research outcomes and cost-efficiency of his placement.

Investment in the student program, including visa, accommodation and airfares, was provided through the GRDC's Statistics for the Australian Grains Industry (SAGI) program, the research team for which is based at the Biometry Hub.

Launched in 2017, phase three of SAGI is a five-year investment. The Southern Node is led by the University of Adelaide, with co-investment from the university and the Southern Australian Research and Development Institute (SARDI) division of Primary Industries and Regions SA (PIRSA).

SAGI South delivers statistical support to about 40 research projects and has a capacity building program for statistical and data analysis skills.

GRDC Research Code ***** Iker Monferrer Master of Science (MSc) Biostatistics Student in Bayesian Variable Selection

More information: Olena Kravchuk, 08 8313 7252, olena.kravchuk@adelaide.edu.au

Aa