Big Data in Plant Science I

Co-organizers: Min Zhang, Professor of Statistics, Department of Statistics, Purdue University; Jianming Yu, Professor and Pioneer Distinguished Chair in Maize Breeding, Department of Agronomy, Iowa State University; Siva Prasad Kumpatla, Global Leader, Data Science, Corteva agriscience, Agriculture division of Dow DuPont

Chair: Min Zhang, Professor of Statistics, Department of Statistics, Purdue University

Speakers

Patrick S. Schnable, C.F. Curtiss Distinguished Professor, Iowa Corn Endowed Chair in Genetics, Director, Plant Sciences Institute, Iowa State University
Alexander E. Lipka, Assistant Professor of Biometry, Department of Crop Sciences, University of Illinois at Urbana-Champaign
Tingting Guo, Postdoctoral Research Associate, Department of Agronomy, Iowa State University
Mitchell R. Tuinstra, Professor of Plant Breeding and Genetics and Wickersham Chair of Excellence in Agricultural Research, Institute for Plant Sciences, Department of Agronomy, Purdue University

Schedule

Thursday, June 7, 10:00 a.m.-12:00 p.m. in STEW 214 AB

Time	Speaker	Title
10:00-10:30 a.m.	Patrick Schnable	The Potential of Predictive Plant Phenotyping to Address (some of) the Challenges Facing Crop Production
	Abstract: To overcome some of the myriad challenges facing sustainable crop production we are seeking to develop statistical models that will predict crop performance in diverse agronomic environments. Crop phenotypes such as yield and drought tolerance are controlled by genotype, environment (considered broadly) and their interaction (GxE). As a consequence of the next generation sequencing revolution genotyping data are now available for a wide diversity of accessions in each of the major crops. The necessary volumes of phenotypic data, however, remain limiting and our understanding of molecular basis of GxE is minimal. To address this limitation, we are collaborating with engineers to construct new sensors and robots to automatically collect large volumes of phenotypic data. New sensors and high-throughput, high-resolution, field-based phenotyping systems will be described. Some of these technologies will be introduced within the context of the Genomes to Fields Initiative.
10:30-11:00 a.m.	Alexander Lipka	Quantification of Non-Additive Genomic Contributions towards Food and Energy-related Crop Traits
	Abstract: Statistical approaches for genome-wide association studies (GWASs) and genomic selection (GS) have enabled the identification of genomic loci associated with agronomically important traits while controlling for false positives, as well as the use of genome-wide marker data to accurately predict trait values. Using these developments as starting points, the Lipka Lab at the University of Illinois is exploring the ability of GWAS and GS approaches to quantify non-additive genomic sources of quantitative trait variability. In this presentation, two different examples of research projects from the Lipka Lab are presented. Collectively, these highlight the impact of genomic properties underlying a trait and species on the performance of statistical approaches for GWAS and GS.
11:00-11:30 a.m.	Tingting Guo	From data mining to knowledge discovery: hidden relationships among genotype, phenotype, and environment
	Tingting Guo¹, Xianran Li¹, Xin Li¹, Qi Mu¹, Randall J. Wisser², Jianming Yu¹ ¹Department of Agronomy, Iowa State University, Ames, IA, USA 50011 ²Department of Plant and Soil Sciences, University of Delaware, Newark, DE, USA 19716 Data mining and knowledge discovery have been attracting a significant amount of attention. In agriculture, enormous data from genomics, phenomics and environmental are generated daily. With the feasibility, effectiveness, and scalability of data mining techniques, rethinking and redesigning the selection and breeding process become pertinent. Changes can be proposed to establish genotype-phenotype relationship so that efficient designs can be made to optimize genomic prediction of hybrid performance. Changes can also be made to establish genotype-phenotype-environment relationship so that hidden patterns and specific factors underlying phenotypic plasticity can be identified and utilized to conduct in-season and on-target performance prediction. To efficiently establish genotype-phenotype relationship, we conducted a multi-specie (maize, wheat, and rice) hybrid performance prediction study. Maize hybrids (276) were generated from diverse founder inbreds, 2,556 wheat hybrids were from an early-stage hybrid breeding system, and 1,439 rice hybrids from an established hybrid breeding system. Patterns of genomic relationships and phenotypic variation were systematically explored from clustering, graphic network analysis, and genetic mating scheme perspective to optimize training set design. Our analysis showed that optimized training set designs significantly outperformed random sampling and other methods that consider either minimizing the prediction error variance (PEV) or maximizing the generalized coefficient of determination (CD). Design optimization and pattern mining are expected to further enhance future studies of complex traits in crops. To explore genotype-phenotype-environment relationship, we conducted a multi-environment trial for a sorghum population with 250 recombinant inbred lines (RILs) across 4 years. A complex flowering time variation across environments, or phenotypic plasticity, was observed. Joint regression analysis (JRA), a typical strategy to understand phenotypic plasticity, was applied to find patterns on phenotype and environment relationship. We upgraded JRA to JGRA by substituting population mean with environment index, and combining it with genomic prediction (GP). The new framework joint genomic regression analysis (JGRA) leverages patterns in genotype, phenotype, and environment, so that mechanisms of phenotypic plasticity can be uncovered, and untested genotypes can be predicted in untested environments. Knowledge discovered from pattern recognition greatly improves our strategies in modern plant breeding.
11:30-12:00 p.m.	Mitch Tuinstra	Plant Breeding in the Omics Era
	Abstract: Our world is expected to grow from seven to more than ten billion people over the next few decades. This increase in population, coupled with changing consumption habits and shifting climates, will create unprecedented demand on food production systems. Significant advances have been made in our understanding of the genes that contribute to crop performance; however, more work is needed to determine how phenotypes or traits emerge from the interaction of genome and environment. To address this question, multidisciplinary research teams representing the Colleges of Agriculture and Engineering and the Purdue Polytechnic at Purdue University are developing aerial and ground-based sensor platforms for growth chamber, greenhouse, and field-based studies of the plant phenome. New sensors and sensor platforms, novel georeferencing techniques, and sophisticated image and data analysis methods (e.g., feature extraction, image segmentation) are being used to quantify variation in plot- and plant-level traits. These measurements provide insights into research plot and field quality, field equipment performance, genotype productivity, physiological plasticity, and spatial variability. Plant breeders are using these and other “omics” tools to address the complex challenges of global food security through collaborative and cross-disciplinary research. New education programs are also being developed to equip a new generation of students with the mindset, skills, and abilities to develop sustainable and climate-proof agriculture production systems.