Skip to main content
ARS Home » Midwest Area » East Lansing, Michigan » Sugarbeet and Bean Research » Research » Research Project #445374

Research Project: Dry Bean Genomic Prediction

Location: Sugarbeet and Bean Research

Project Number: 5050-21430-012-012-A
Project Type: Cooperative Agreement

Start Date: Dec 18, 2023
End Date: Mar 31, 2027

Implement genomic-assisted breeding tools to jointly predict top yielding and canning quality breeding lines in two dry bean market classes.

Genomic prediction for seed yield and canning quality will be implemented in cooperator navy and black bean breeding germplasm for cultivar improvement. Proposed multi-environment and multi-trait genomic selection index (GSI) strategy implemented in the cooperator dry bean breeding program. This GSI strategy will use multiple independent PYT trials grown in 2021-2024 to be used as the training set. This training set will be used to predict 2022-2023 AYT black and navy breeding lines in 2024 and a set of selected breeding lines again in 2025 during regional testing for yield and canning quality using all environmental data to improve the accuracy of the models and accelerate the development of two dry bean cultivars. PYT: Preliminary Yield Trials; AYT: Advanced Yield Trials. Genomic prediction models and cross-validation We will evaluate several prediction models including single trait, multi-trait, models including genotype x environment interaction, and multi-trait multi-environment genomic prediction model to predict grain yield and canning quality. In addition, we will develop a genomic selection index(GSI) method to select the best genotypes based on their net merit derived of multiple traits in a sparse testing approach. Different cross-validation schemes that mimic actual scenarios in the breeding program will be used to estimate the prediction accuracy of these traits and compare the performance of different models. 1.8.1 Single-trait model: Two parametric models will be considered. The first one is based on the main effects of the marker SNPs via covariance structures (GBLUP, VanRaden 2008) while the second model will return specific genomic effects for each environment by allowing the interaction between marker and environments (reaction norm model, Jarquin et al., 2014). Both models will be fitted using the BGLR package in R. ARS role: provide the phenotypic and genomic data for the models.