Determination of the optimal number of markers and individuals in a training population necessary for maximum prediction accuracy in F₂ populations by using genomic selection models

Abstract

ABSTRACT. Genomic selection is a useful technique to assist breeders in selecting the best genotypes accurately. Phenotypic selection in the F 2 generation presents with low accuracy as each genotype is represented by one individual; thus, genomic selection can increase selection accuracy at this stage of the breeding program. This study aimed to establish the optimal number of individuals required to compose the training population and to establish the amount of markers necessary to obtain the maximum accuracy by genomic selection methods in F 2 populations. F 2 populations with 1000 individuals were simulated, and six traits were simulated with different heritability values (5, 20, 40, 60, 80 and 99%). Ridge regression best linear unbiased prediction was used in all analyses. Genomic selection models were set by varying the number of individuals in the training population (2 to 1000 individuals) and markers (2 to 3060 markers). Phenotypic accuracy, genotypic accuracy, genetic variance, residual variance, and heritability were evaluated. Greater the number of individuals in the training population, higher was the accuracy; the values of genotypic and residual variances and heritability were close to the optimum value. Higher the heritability of the trait, higher is the number of markers necessary to obtain maximum accuracy, ranging from 200 for the trait with 5% heritability to 900 for the trait with 99% heritability. Therefore, genomic selection models for prediction in F 2 populations must consist of 200 to 900 markers of major effect on the trait and more than 600 individuals in the training population.

Training Population Optimization for Prediction of Cassava Brown Streak Disease Resistance in West African Clones

“…In the present study, prediction accuracies for CBSD3s and CBSDRs were 8% and 18% higher for high-density (WGS-imputed) markers than low-density (GBS) markers from single kernel G-BLUP model for the optimized training population of size of 200 clones (Table S11). Several studies have demonstrated increased prediction accuracies as a function of increase marker density (Peixoto et al , 2016; Wang et al , 2017). In a recent study, using NaCRRI training population, prediction for CBSD-related traits, in a single kernel G-BLUP model was not improved by whole-genome imputation (Lozano et al , 2017).…”

Multi-trait genomic selection indexes applied to identification of superior genotypes

“…Each linkage group was simulated with 100 cM, comprising 100 codominant molecular markers, spaced equidistantly (1 cM), totaling 1000 markers. The number of markers used in the simulation is considered adequate for the study purpose and is similar with other simulations studies (Oliveira et al 2021;Peixoto et al 2016, Sant' Anna et al 2019.…”

“…Bhering et al (2015) and Spindel et al (2015) showed that a small number of markers can be used for high precision genomic selection for many traits. Peixoto et al (2016), in a study with simulated data, concluded that a genomic selection model that uses 300-800 markers is sufficient to capture all the genetic variance, and to decrease the residual variance, to obtain the maximum prediction precision of an F 2 population.…”

Breeding Jatropha curcas by genomic selection: A pilot assessment of the accuracy of predictive models

“…However, we could not do it because we had problems with the DNA extraction, and just a few plants could be evaluated. When we estimated these parameters using all plants in the experiment this problem was solved as showed by Junqueira et al [ 25 ] and Peixoto et al [ 41 ].…”