Criteria for evaluating molecular markers: Comprehensive quality metrics to improve marker-assisted selection

• DOI: 10.1371/journal.pone.0210529
• Publication Date: 2019-01-15T18:29:32Z
• Authors: John Damien Platten; Joshua Nathaniel ...; Rochelle E. Zantua
• Keywords: Genetic Markers; Quantitative trait locus; 0301 basic medicine; Marker-Assisted Selection; Science; Quantitative Trait Loci; Population; Trait; Set (abstract data type); Plant Science; Molecular marker; Gene; Quantum mechanics; Agricultural and Biological Sciences; Computational biology; 03 medical and health sciences; Selection (genetic algorithm); Cultivar Evaluation and Mega-Environment Investigation; Biochemistry, Genetics and Molecular Biology; Genetic Diversity and Breeding of Wheat; Machine learning; Genetics; Biomass; Selection, Genetic; Genetic marker; Biology; 0303 health sciences; Physics; Q; R; Chromosome Mapping; Reproducibility of Results; Life Sciences; QTL Mapping; Power (physics); Computer science; Programming language; Plant Breeding; Genetics, Population; Reliability (semiconductor); Environmental health; Genetic Architecture of Quantitative Traits; FOS: Biological sciences; Medicine; Marker-assisted selection; Research Article; Microsatellite Repeats

AbstractDespite strong interest over many years, the usage of quantitative trait loci in plant breeding has often failed to live up to expectations. A key weak point in the utilisation of QTLs is the “quality” of markers used during marker-assisted selection (MAS): unreliable markers result in variable outcomes, leading to a perception that MAS products fail to achieve reliable improvement. Most reports of markers used for MAS focus on markers derived from the mapping population. There are very few studies that examine the reliability of these markers in other genetic backgrounds, and critically, no metrics exist to describe and quantify this reliability. To improve the MAS process, this work proposes five core metrics that fully describe the reliability of a marker. These metrics give a comprehensive and quantitative measure of the ability of a marker to correctly classify germplasm as QTL[+]/[-], particularly against a background of high allelic diversity. Markers that score well on these metrics will have far higher reliability in breeding, and deficiencies in specific metrics give information on circumstances under which a marker may not be reliable. The metrics are applicable across different marker types and platforms, allowing an objective comparison of the performance of different markers irrespective of the platform. Evaluating markers using these metrics demonstrates that trait-specific markers consistently out-perform markers designed for other purposes. These metrics also provide a superb set of criteria for designing superior marker systems for a target QTL, enabling the selection of an optimal marker set before committing to design.