Location: Dale Bumpers National Rice Research Center
Title: Genomic prediction and QTL mapping of root system architecture and above-ground agronomic traits in rice (Oryza sativa L.) with a multi-trait index and Bayesian networksAuthor
Sharma, Santosh | |
Pinson, Shannon | |
Gealy, David | |
Edwards, Jeremy |
Submitted to: G3, Genes/Genomes/Genetics
Publication Type: Peer Reviewed Journal Publication Acceptance Date: 2/17/2021 Publication Date: 5/28/2021 Citation: Sharma, S., Pinson, S.R., Gealy, D.R., Edwards, J. 2021. Genomic prediction and QTL mapping of root system architecture and above-ground agronomic traits in rice (Oryza sativa L.) with a multi-trait index and Bayesian networks. G3, Genes/Genomes/Genetics. https://doi.org/10.1093/g3journal/jkab178. DOI: https://doi.org/10.1093/g3journal/jkab178 Interpretive Summary: Roots allow plants to acquire critically necessary water and nutrients, leading to the expectation that a vigorous root system would be essential for the development of a high yielding crop variety. Largely due to the difficulty in studying roots in the field, we know very little about how specific differences in root structure or “architecture” contribute to above-ground shoot growth and grain yield. Also, because of the difficulty in evaluating roots there is a need to develop novel tools that allow breeders to make root-based selections without having to actually observe the roots of hundreds to thousands of breeding progeny. Marker-assisted-selection can allow breeders to use genetic markers in place of actual phenotyping to identify progeny lines that contain desired genes or gene combinations. Before marker-assisted-selection can be used, however, one must have prior knowledge as to what particular traits and genes are desired, and what markers can be used to identify them. In this study, we used two gene mapping analysis methods to discover marker-associations with root structure attributes predicted to impact overall plant growth and yield. This allowed us to compare the mapping strategies as well as identify more genes affecting the various root structures. We first identified genes using data on individual traits, then used a multi-trait machine learning model (a Bayesian network) that allowed us to consider trait-to-trait associations as we sought to find marker-trait associations. Genes identified using the network model were able to explain how roots affect above-ground growth, while the genes identified using the individual traits were not only fewer in number, but also less informative of above-ground health. Genomic selection is an extension of marker-assisted-selection that makes use of markers along the entire genome to predict the overall breeding value, or optimum gene content at numerous genes whose specific trait effects need not be known before the study and breeding selections can be made. To make genomic predictions of breeding value, one first collects genetic and trait data on a portion or subset of the breeding population, then uses that data to create mathematical genetic models that can then be used to predict the breeding value of all the remaining progeny based only on genotypic data. We conducted a prototype study to evaluate the ability of these genetic models to impart breeding progress by using data on 13 root traits measured from a subset of 69 progeny. We created a model that predicted those traits for 204 breeding progeny on which we had genetic but no root trait data. We then created and used for the first time a genomic selection rank sums index (GSRI) to merge those 13 predicted root traits into one single multi-trait estimate of overall root value. We used the GSRI to select two groups of progeny in opposite directions (high vs low GSRI), which we then phenotyped. When the root traits were compared in the two groups, we found that making selections using merged genomic predictions was successful resulting in opposite shifts in average root traits between the progeny selected for high versus low GSRI. The cost of obtaining genotypic data has rapidly decreased in the past decade unlike the cost of labor for phenotyping. This has increased interest in adopting methods that reduce the amount of phenotyping needed to achieve breeding progress. Our study demonstrates an analysis approach that can be used by breeders to simultaneously select for numerous below-ground and above-ground traits using genomic information. Technical Abstract: Root system architecture (RSA) is a crucial factor in resource acquisition and plant productivity. Roots are difficult to phenotype in the field, thus new tools for predicting phenotype from genotype are particularly valuable for plant breeders aiming to improve RSA. This study identifies quantitative trait loci (QTLs) for RSA and agronomic traits in a rice (Oryza sativa) recombinant inbred line (RIL) population derived from parents with contrasting RSA traits (PI312777 x Katy). The lines were phenotyped for agronomic traits in the field, and separately grown as seedlings on agar plates which were imaged to extract RSA trait measurements. QTLs were discovered from conventional linkage analysis and from a machine learning approach using a Bayesian network (BN) consisting of genome-wide SNP data and phenotypic data. The genomic prediction accuracies (GPAs) of multi-QTL models and the BN analysis were compared with the several standard genomic prediction methods. We found GPAs were improved using multi-trait (BN) compared to single trait genomic prediction in traits with low to moderate heritability. Two groups of individuals were selected based on genomic predictions and a modified rank sum index (GSRI) indicating their divergence across multiple RSA traits. Selections made on genomic predictions did result in differences between the group means for numerous RSA. The ranking accuracy across RSA traits among the individual selected RILs ranged from 0.14 for root volume to 0.59 for lateral root tips. We conclude that the multi-trait genomic prediction model using BN can in some cases improve the GPA of RSA and agronomic traits, and the GSRI approach is useful to simultaneously select for a desired set of RSA traits in a segregating population. |