Multiparental QTL analysis: can we do it in polyploids?

Alejandro Thérèse Navarro is a first-year PhD student in the Laboratory of Plant Breeding at Wageningen University & Research, The Netherlands, working on his thesis titled “Molecular breeding and evolution in allopolyploids: novel and applied methodologies.” He focuses on statistical tool development for the analysis of polyploid crops. His main interest is computational analysis of plant breeding data in order to understand the biological characteristics of crops. To that end, he has been studying polyploid genetic mapping. He obtained his MSc on Plant Biotechnology at the same laboratory with his thesis research: “QTL mapping in Multiparental Polyploid Populations: Development of Computational Methods in R”. During his PhD he continues to work on this topic. Polyploid quantitative trait locus (QTL) analysis tools have been developed in recent years and remain limited in the population types they can accept. They tend to focus either on biparental crosses or diversity panels for genome-wide association studies (GWAS). The former is a limited tool due to the restricted genetic diversity of the cross, while with the latter, weak or rare alleles cannot be estimated. Multiparental populations (MPPs) are an alternative where a few parents are crossed to obtain connected biparental populations. In plant breeding efforts, such schemes are common, and thus developing QTL analysis for MPPs increases the usefulness of these ad hoc MPPs. In this study, we simulated tetraploid MPPs and developed a QTL modelling approach to find the simulated QTLs. This allowed us to identify the main factors influencing QTL analysis in polyploid MPPs. We present the QTL modelling approach that best suits polyploid MPPs, based on identity-by-descent (IBD) QTL models presented in previous literature. These models require identification of the number of alleles segregating in the population. For that purpose, haplotyping, the concatenation of multiple SNP alleles, can be used. Thus, biallelic SNPs can be transformed into multiallelic markers, to better estimate the QTL effects. However, polyploid haplotyping is a complex problem, and although it has received attention recently, no methods have been developed yet for polyploid MPPs. Additionally, modelling approaches for MPPs must accommodate genetic structure, the non-homogeneous distribution of genetic similarity across the population. Such models include mixed models (such as the “unified mixed model”) and Bayesian models (such as those implemented for diploids in FlexQTL). To conclude, our results showed that the correlation between genetic structure and phenotype variation decreases QTL detection power in MPPs. That is to say, if all genetically similar parents contribute similar genetic effects, QTL peaks cannot be detected. Moreover, our results seem to indicate that the haplotype-based approach is only slightly more sensitive and accurate than the SNP-based approach. As the haplotype-based approach is computationally more demanding, SNP-based QTL analysis for QTL detection is recommended, followed by haplotype-based refinement.

Alejandro Thérèse Navarro won an ISHS Young Minds Award for the best oral presentation at the XXVI International Eucarpia Symposium Section Ornamentals: Editing Novelty in Germany in September 2019.

Alejandro Thérèse Navarro, Laboratory of Plant Breeding, Droevendaalsesteeg 1, office E2.111, 6708 PG, Wageningen University & Research, Wageningen, The Netherlands, e-mail: alejandro.theresenavarro@wur.nl

The article is available in Chronica Horticulturae