Premium
Perspectives for Genomic Selection Applications and Research in Plants
Author(s) -
Heslot Nicolas,
Jannink JeanLuc,
Sorrells Mark E.
Publication year - 2015
Publication title -
crop science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.76
H-Index - 147
eISSN - 1435-0653
pISSN - 0011-183X
DOI - 10.2135/cropsci2014.03.0249
Subject(s) - selection (genetic algorithm) , genomic selection , biology , trait , resource (disambiguation) , population , genetic gain , quantitative trait locus , microbiology and biotechnology , plant breeding , marker assisted selection , asset (computer security) , computational biology , computer science , genetics , machine learning , genotype , genetic variation , gene , single nucleotide polymorphism , computer network , demography , computer security , sociology , agronomy , programming language
Genomic selection (GS) has created a lot of excitement and expectations in the animal‐ and plant‐breeding research communities. In this review, we briefly describe how genomic prediction can be integrated into breeding efforts and point out achievements and areas where more research is needed. Genomic selection provides many opportunities to increase genetic gain in plant breeding per unit time and cost. Early empirical and simulation results are promising, but for GS to deliver genetic gains, careful consideration of the problem of optimal resource allocation is needed. Consideration of the cost‐benefit balance of using markers for each trait and stage of the breeding cycle is needed, moving beyond only focusing on recurrent selection with GS on a few complex traits, using prediction on unphenotyped individuals. With decreasing marker cost, phenotype data is quickly becoming the most valuable asset and marker‐assisted selection strategies should focus on making the most of scarce and expensive phenotypes. It is important to realize that markers can also improve accuracy of selection for phenotyped individuals. Use of markers as an aid to phenotype analysis suggests a number of new strategies in terms of experimental design and multi‐trait models. GS also provides new ways to analyze and deal with genotype by environment interactions. Lastly, we point to some recent results showing that new models are needed to improve predictions particularly with respect to the use of distantly related individuals in the training population.