Computer Simulation in Plant Breeding Xin Li1, Chengsong Zhu1, Jiankang Wang2, and Jianming Yu1 1Department of Agronomy, Kansas State University, Manhattan, KS, USA 2Institute of Crop Science and CIMMYT China, Chinese Academy of Agricultural Sciences, Beijing, China Introduction Application I: Breeding Method • As a bridge between theory and experimentation, computer simulation has become a powerful tool in scientific research. It provides not only preliminary validation of theories, but also guidelines for empirical experiments. • Plant breeding is to develop superior genotypes with available genetic and non-genetic resources, during which selecting the best breeding strategy would maximize genetic gain and achieve costeffectiveness. • Computer simulation can establish the breeding process in silico and identify candidates of the optimum combination of various factors, which can then be validated empirically. Insights gained from empirical studies, in turn, can be further incorporated into computer simulations. Computer simulation can be employed to compare different breeding strategies, incorporating various factors simultaneously, such as gene information, cross scheme, propagation method, population size, selection intensity, and number of generations. Thus, we can use computer simulation to decide which breeding strategy could lead to higher selection gain. b F1 a F2 F1 ::: F2 ........................ F3 ........................ F4 ........................ 4.2 4.2 3.8 3.8 20 QTL MARS 20 QTL GS 100 QTL MARS 100 QTL GS 3.4 F5 128 128 256 256 markers markers markers markers 48 96 48 96 3.4 F7 F8 3.0 Phenotype value Computer simulation Plant breeding 3.0 32 64 128 256 512 768 288 Number of markers used in two selection cycles c 576 1152 Number of plants used in genomic selection 90 Fig. 3 An Example of Computer Simulation in Breeding Method Research Genotype value value Genotype (%(%of of maximum) max) QU-GENE engine Trait value APSIM Soil Water CO2 Nitrogen Radiation Temperature Fig. 5 An Example of Computer Simulation in Crop Modeling Research Application IV: Crop Modeling DHs DHs DHs DHs F6 80 Comparison of Marker Assisted Recurrent Selection (MARS) and Genomic Selection (GS) 2 H=0.1 H =0.1ES ES 2 H=0.1 H =0.1SSD SSD 2 H=0.5 H =0.5ES ES 2 H=0.5 H =0.5SSD SSD 60 Gene number Gene effect Epistasis Environment GEI Yield value P1×P2 P2 70 Population Linking QU-GENE with APSIM (Agricultural Production Systems sIMulator) b Response Density P1 Vanoeveren, A. J., & Stam, P. 1992. Heredity, 69, 342-351. Tanksley, S. D., & Nelson, J. C. 1996. Theoretical and Applied Genetics, 92, 191203. Meuwissen, T. H. E. et al. 2001. Genetics, 157, 1819-1829. Bernardo, R., & Yu, J. 2007. Crop Science, 47, 1082-1090. Response a Selection in MET Application II: Gene Mapping • Computer simulation can integrate crop physiological models, environmental information, and genetic compositions of different crops to fill the gap between genotype and phenotype. • We can use computer simulation to predict the performance of different cultivars in the target population of environments, thus facilitate the plant breeding process. • When coupled with climate simulation models, crop models can be used to predict the possible influences from climate change on crop production, which can subsequently provide guidelines for plant breeding. Chapman, S. et al. 2003. Agronomy Journal, 95, 99-113. Yin, X. Y. et al. 2005. Journal of Experimental Botany, 56, 967-976. Hodson, D., & White, J. 2010. Climate Change and Crop Production, 245-262. 50 200 300 400 plants Number of FF22 plants Comparative simulation of ES and SSD Fig. 1 Joining Computer Simulation with Plant Breeding For example, computer simulation can be used to compare two breeding methods, Early Selection (ES) and Single Seed Descent (SSD) Outline In this review, we discussed the application of computer simulation in different aspects of plant breeding. First, we briefly summarized the history of plant breeding and computer simulation, and how computer simulation can be used to facilitate the breeding process. Next, we partitioned the utility of computer simulation into different research areas of plant breeding, including breeding method comparison, gene mapping, genetic modeling, and crop modeling. Then we discussed computational issues involved in the simulation process. Finally, the application of computer simulation in the future was discussed. • Computer simulation can be applied to gene mapping study to validate the effectiveness of new mapping methods or assess the factors influencing mapping power, such as population type and size, marker number and density, heritability, and number of QTL. • Computer simulation can also help us determine the significant threshold (LOD score) and confidence interval, which otherwise are difficult to obtain. Lander, E. S., & Botstein, D. 1989. Genetics, 121, 185-199. Churchill, G. A., & Doerge, R. W. 1994. Genetics, 138, 963-971. Zeng, Z. B. 1994. Genetics, 136, 1457-1468. Beavis, W. D. 1998. Molecular Dissection of Complex Traits, 145-162. Yu, J. et al. 2006. Nature Genetics, 38, 203-208. a b 100 Estimated-simulated additive effects 100 2 10 QTL h =0.3 2 10 QTL h =0.95 2 40 QTL h =0.3 2 40 QTL h =0.95 80 Power (%) 0 60 40 Breeding Method • • Compare breeding strategy Assess factors influencing marker assisted selection 20 0 100 500 Size of mapping population Gene Mapping Crop Modeling • Assess factors influencing mapping power • Determine significant threshold and confidence interval of QTL position Computer Simulation in Plant Breeding • • Use gene information as model parameters Predict crop performance in target environments Genetic Modeling • Combine genetic and gene by environment interaction to simulate the whole plant breeding process Fig. 2 Various Applications of Computer Simulation in Plant Breeding 1000 Perspectives • Research in establishing genotype-phenotype relationship, and developing new breeding methods, have been proposed as key factors to realize the potential brought by ultrahigh throughput genomic technologies in plant breeding, and computer simulation, undoubtedly, will play a key role in this process. • As a tool to aid decision making and resource allocation, computer simulation would undertake the responsibility of transferring the experimental outcome from laboratory to realistic agriculture production, predicting the outcome of breeding decision, directing gene mapping, and tackling genotype by environment interaction and climate change. 2 10 QTL h =0.3 2 10 QTL h =0.95 2 40 QTL h =0.3 2 40 QTL h =0.95 3 2 1 0 100 500 1000 Size of mapping population Fig. 4 An Example of Computer Simulation in Gene Mapping Research Effects of heritability and sample size on the power, precision and accuracy of QTL study Application III: Genetic Modeling Plant breeding simulation platforms are potent tools which can simulate the whole plant breeding process. They use genetic and geneenvironment interaction information to assist in decision making, e.g. predicting cross performance and comparing selection methods. Podlich, D. W., & Cooper, M. 1998. Bioinformatics, 14, 632-653. Wang, J. K. et al. 2004. Crop Science, 44, 2006-2018. Key References Allard, R. W. 1960. Principles of plant breeding. Falconer, D. S., & Mackay, T. F. C. 1996. Introduction to quantitative genetics. Hartl, D. L., & Clark, A. G. 1997. Principles of population genetics. Lynch, M., & Walsh, B. 1997. Genetics and analysis of quantitative traits. Mackay, T. F. C. 2001. The genetic architecture of quantitative traits. Annual Review of Genetics, 35, 303-339. Bernardo, R. 2002. Breeding for quantitative traits in plants. Doerge, R. W. 2002. Mapping and analysis of quantitative trait loci in experimental populations. Nature Reviews Genetics, 3, 43-52. Holland, J. B. 2007. Genetic architecture of complex traits in plants. Current Opinion in Plant Biology, 10, 156-161. Acknowledgements This work was supported by the Plant Feedstock Genomics Program of USDA and DOE, the Plant Genome Program of NSF, the Targeted Excellence Program of Kansas State University, and the Great Plains Sorghum Improvement and Utilization Center.