Optimization of Selective Phenotyping and Population Design for Genomic Prediction
- PDF / 680,134 Bytes
- 22 Pages / 496.063 x 720 pts Page_size
- 30 Downloads / 208 Views
1. INTRODUCTION Genomic prediction (Meuwissen et al. 2001), the use of high density whole genome markers to predict performance of both observed and unobserved individuals has fostered great interest in the plant breeding community (Heslot et al. 2015; Auinger et al. 2016; Bernardo 2016; Marulanda et al. 2016). A critical feature of any genomic prediction strategy is the population design (Marulanda et al. 2015; Wurschum et al. 2017). This opens several new design questions such as how to select a subset of preexisting individuals for phenotyping based on the molecular marker data to estimate marker effects with the highest precision “selective phenotyping”. Marker data might be available on a large number of individuals but budget or logistical constraints prevent phenotyping of the whole population
N. Heslot (B) and V. Feoktistov, Biostatistics Department, Limagrain Field Seeds Research, Chappes Research Center, Chappes, France (E-mail: [email protected]) . © 2020 International Biometric Society Journal of Agricultural, Biological, and Environmental Statistics https://doi.org/10.1007/s13253-020-00415-1
N. Heslot, V. Feoktistov
(Yu et al. 2016). Methods for optimal selection of individuals to phenotype to best predict the unobserved are then needed. Mixed model based criteria (Laloë 1993) were proposed to optimize the choice of individuals to phenotype using as an input the expected trait heritability and markers or pedigree data (Berro et al. 2019; Bustos-Korts et al. 2016; Isidro et al. 2015; Rincent et al. 2012). This design problem might benefit from a more sophisticated optimization method. Such optimization is a part of a broader class of so called “modelbased design” also used to optimize field trials design (Butler et al. 2014; Feoktistov et al. 2017; Williams et al. 2014). Similarly, in hybrid breeding, male and female inbreds are usually available with high density markers and what is of interest is the identification of high performing male and female combinations. In most cases, it is unfeasible to field test all of the possible hybrid combinations. Molecular markers can be used here as well to predict unobserved hybrid combinations, often referred as hybrid prediction (Albrecht et al. 2014). How to choose the hybrids combinations to create and phenotype to best predict the performance of the unobserved hybrid combinations “hybrid crossing plan” is another important design question. Approaches proposed for the “selective phenotyping” design question have not been tested to solve this one. One might expect a large impact of such design optimization as usually only a very small fractions of all the possible hybrid combinations can be field tested. Last given a list of individuals to be used to create new individuals and logistical constraints, which crosses to make and how many individuals per crosses to produce to optimize marker effects estimation with a budget constraint “connected crosses design”. A method is needed, using the potential parents marker profile, to evaluate the expected
Data Loading...
