Using public databases for genomic prediction of tropical maize lines

Abstract

In this paper, the aims were (a) to test the usefulness of using genomic and phenotypic information from public databases (open access) to predict genetic values for tropical maize inbred lines regarding plant and ear height; (b) to identify how the population structure, the use of optimized training sets (OTSs) and the amount of information originating from public databases affect the predictive ability. Thus, 29 training sets (TSs) were defined considering three diversity panels: the University of São Paulo (USP—validation set (VS)) and the ASSO and USDA North Central Regional Plant Introduction Station (NCRPIS) (external public panels—predictors), which were divided into four scenarios with different TS configurations. We showed that it is possible to use public datasets as a primary TS and that population structure can modify the predictive abilities of GS. In the four scenarios proposed, very large or very small sets did not provide predictive abilities over 0.53 for GS. However, OTSs composed of 250 individuals were sufficient to achieve predictive abilities over this limit.

Enviromic-based Kernels Optimize Resource Allocation with Multi-trait Multi-environment Genomic Prediction for Tropical Maize

“…Moreover, the samples have a good distribution of genotypes, including those genotypes that perform well, and those that are not so good, bringing positive impacts on PA (Michel et al, 2020). As expected and similar to what Pinho Morais et al (2020) found, with a small effective population size, PA is diminished, since the sample contains small genetic variability.…”

“…The establishment of the TRN, which should be representative in terms of size, diversity, and the relationship of the individuals to be predicted, is the key to success in GS (Jannink et al, 2010;Akdemir et al, 2015;Crossa et al, 2017;Varshney, 2017;Ibba et al, 2020). For that, the main objectives are to minimize costs associated with phenotyping by selecting smaller training populations, and maximize the predictive ability for the individuals of the TST through efficient resource allocation (Isidro et al, 2015;Lado et al, 2018;Pinho Morais et al, 2020;Riedelsheimer & Melchinger, 2013;Technow et al, 2014). Additionally, there is a lack of knowledge on how to distribute genotypes optimally in multi-environment trials in order to achieve the best balance between the number of genotypes tested in the field and the predictive capacity of GP models, and maximize the selection gain with fixed area and budget resources (Jarquin et al, 2020).…”

“…In this context, many studies have been carried out aiming to establishing the balance between investment and efficiency through different methods, experiment design, statistical analysis, and TRN composition. For instance, the genetic algorithm to design training populations developed by Akdemir (2017) was tested by Pinho Morais et al (2020) for several population sizes. The responses were compared with randomly selected populations, noting that optimizing TRN can be effective to obtain satisfactory accuracies.…”

Optimizing Genomic-Enabled Prediction in Small-Scale Maize Hybrid Breeding Programs: A Roadmap Review

“…Due to the scarcity of resources in the initial phases, we addressed the possibility of incorporating public databases in the composition of our training populations (Morais et al, 2020). Small-scale public and private programs with limited budgets often lack the financial ability to genotyping a considerable number of individuals to apply GP efficiently.…”

“…Small-scale public and private programs with limited budgets often lack the financial ability to genotyping a considerable number of individuals to apply GP efficiently. In this regard, Morais et al (2020) These databases contained phenotypic information regarding plant height (PH, in cm), ear height (EH, in cm), and the SNP markers data. A total of 29 training populations (TPs) were defined and divided into four scenarios to determine the best strategy to apply public databases to predict lines.…”