Prediction-based normalization for developmental heterochrony in parallel molecular-level and phenomic studies in plants

• Applicant: Dr. Kerstin Neumann
• Subject Area: Plant Physiology ; General Genetics and Functional Genome Biology ; Bioinformatics and Theoretical Biology ; Plant Cultivation, Plant Nutrition, Agricultural Technology
• Term: since 2023
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 504808160

The transient nature of all developmental and signalling processes in plants requires very specific planning of the sampling time and adjustment of that time between genotypes differing in development. Surprisingly, no established methods for a dynamic planning for sampling in genomic experiments, nor tools for real-time monitoring of developmental stages between diverse genotypes exist. Here, we will establish statistical and machine learning methods for monitoring of plant developmental stages using a high throughput imaging system and deliver tools for dynamic planning of sampling in large-scale genomic experiments. Our main hypothesis is that by adjusting the experimental plans to the shifts in development of plants actually observed in the trials we will enable higher statistical power of linking phenotypes with molecular parameters and improve the efficiency of estimation of genetic effects. We call the phenomenon of different reactions of plants to environmental signals, resulting in entering developmental stages at different time points, a (developmental) heterochrony. Our main research question is: how phenotypic and environmental data processed and shared with the experimenter in real time can serve as the input to the decision support system allowing for dynamic planning of sampling for molecular-level assays in the presence of developmental heterochrony. We will propose methods of data processing and integration necessary for dynamic experiment planning using two statistical approaches: multivariate linear mixed models and functional data analysis. We will achieve real time classification of plant developmental stages using deep learning-based image analysis. We will implement the statistical and machine learning-based methods in real time as a support tool for planning decisions in crop genomics experiments. We will verify the developed algorithms for the analysis of datasets already available to partners from a range of previous experiments in diverse wheat and barley panels. We will also validate and further optimize the methods on the basis of newly generated molecular-level and phenomics data. The project will have a significant impact on quantitative genetics, crops breeding and phenotypic data analysis, as it addresses important but so far largely overlooked aspects of experimental design, data modeling and phenotyping data infrastructure development. Last but not least, proposed proof-of-concept experiments will shed a new light on temporal aspects of barley drought tolerance and provide new unique data on the interaction between genetic determinants of stress tolerance, plant development and the timing of stress occurrence.

• DFG Programme: Research Grants
• International Connection: Poland
• Cooperation Partner: Professor Pawel Krajewski, Ph.D.