Statisticians develop new methods for analyzing high-throughput, non-invasive phenotyping data

High-throughput, noninvasive phenotyping is a technology that is having a revolutionary impact on plant research. A recent publication in Nature Communications describes its use in the controlled environment of a glasshouse as part of a rice experiment carried out by a team consisting of researchers from the King Abdullah University of Science and Technology (KAUST), the University of South Australia (UniSA) and The Plant Accelerator®, Australian Plant Phenomics Facility.

In the experiment, two panels, each of more than 250 of rice varieties, were grown under low and high salinity conditions with a view to identifying new genetic loci that impart salt tolerance. During the experiment the growth of each plant is measured daily by using images taken with digital cameras to monitor biomass and shoot development.

The result is a large data set (15,000 values for a single trait) that is complicated by involving observations over time and which is difficult to analyse with traditional methods. Many researchers prefer to use growth functions to analyse growth data such as this. However, Dr Chris Brien, of UniSA’s Phenomics and Bioinformatics Research Centre, makes the point that ‘growth does not always conform to a particular pre-determined equation’. Dr Brien developed a computationally efficient method of analyzing the data that ‘allows plants to express their individuality’. It is based on the well-established statistical techniques of spline fitting and mixed modelling. The novelty lies in the way that these techniques are deployed.

Dr Helena Oakey (KAUST) formulated a new GWAS model and was able to show that some genes, for example those connected with signalling processes, were important to plant growth in the first two to six days after salt application, while other genes became prominent later.

It is anticipated that the ability to examine the dynamics of genetic response as described in the paper will be a major benefit to researchers conducting breeding experiments.

Find the full article:

Obtain the data and scripts at or the R package imageData for the phenotypic analysis CRAN Package imageData

Dr Chris Brien
Adjunct Senior Lecturer, Phenomics and Bioinformatics Research Centre, The University of South Australia
Senior Biostatistician, The Plant Accelerator, Australian Plant Phenomics Facility, University of Adelaide,,

Areas of study and research

+ Click to minimise