Resources and statistical approaches for mapping determinants of trait of interest and genomic selection 

Association panels, genotyping, statistical methods

WP leader: Laurence Moreau (INRA UMR GQE)
Private partners involved: Biogemma, Caussade Semences, Euralis Semences, KWS, Limagrain, Maïsadour Semences, Syngenta, RAGT 2n.


WP4 objective is to provide optimized resources (plant material, molecular information) and statistical approaches to carry out genome-wide association (GWA) and genome-wide selection (GWS) using phenotypic data obtained in WP5. This work rely on WP3 results and aims at developing data and tools for WP5 and WP8

WP4 is organized into 5 complementary tasks:

Task 4.0 : Panel assembly and seed production – Leader: Cyril Bauland (INRA UMR GQE).
Panel assembly is crucial for association mapping and genomic selection. The objectives of this task are to assemble and produce seeds for (i) line panels (for the Dent and Flint pool) and (ii) single-cross hybrid panels adapted for studying hybrid value and heterosis.
A Dent panel of 349 lines was assembled and evaluated for test-cross value in WP5. The assembly of a Flint panel including lines from T4.2 is ongoing. Two “heterosis” panels have been assembled: one is a factorial between Dent and Flint lines and the second is a partial diallel between admixed Dent-Flint lines developed in T4.2.

T4.1 : Analysis of diversity available in French collections of lines and open pollinated varieties (OPVs) – Leader: B. Gouesnard (INRA UMR AGAP)
Large collections of genetic resources have been assembled and are conserved at INRA for inbred lines and OPVs. These collections have been only partially characterized phenotypically and genetically. The objectives of this task are to characterize these collections, focusing on the Flint group, in order to better understand diversity organization and identify original sources of diversity in view of integrating them in new panels (T4.0 and T4.2).
A collection of more than 1000 inbred lines was characterized by genotyping by sequencing (GBS) and evaluated for per se value for traits related to phenology and ear morphology. Results illustrate the originality of the Flint material compared to USDA collection, enabled GWAS detection of loci involved in the studied traits (Gouesnard et al. TAG 2017) and the identification of original sources of diversity. Flint populations were phenotyped and 127 of them were genotyped (50k SNP) which revealed an Eastern-Western diversity gradient along the Pyrenean region. Further analyses are currently done.

Task 4.2: New inbred line development – Leaders: A. Charcosset et C. Bauland (INRA UMR GQE)
Panel assembly is limited by the genetic diversity available within an earliness window compatible with a relevant phenotypic evaluation and the poor agronomic value of some accessions (lodging susceptibility of the Flint material). Moreover, the advent of population structure in panels limits the power of GWAS and the transfer of genomic predictions from one group to the other.
The objectives are to create original inbred lines to improve GWAS power in the Amaizing panels. A first set of Dent-Flint lines was created by doubled-haploidization (DH) from Dent-Flint hybrids. These lines will make it possible to evaluate the interest of admixed lines in GS. A partial diallel between them (T4.2) will be used to identify loci involved in heterosis phenomenon (WP5). A second set of DH lines was derived from crosses between Flint lines identified in T4.1. Phenotypic and molecular evaluation of these lines will be used to identify the ones that will be included in the new Flint panel (T4.0) and perform GWAS for per-se value.

Task 4.3 : Genotyping of panels  Leader : S. Nicolas (INRA UMR GQE)
Developments of molecular biology and resequencing approaches opened the way to a characterization of panels for a high density of different types of polymorphisms (SNP, presence absent variants (PAV), epigenetic marks…). For cost or technical issues it is nevertheless not always possible to characterize a whole panel. It is therefore important to optimize strategies and tools for imputing missing data.
Task objectives are to provide partners with dense genotyping data for implementing GWAS and GS in WP5. A matrix of 1M SNP was assembled for the Dent panel from genotyping issued from 50k and 600k SNP arrays and GBS. This panel will be genotyped for PAV using an Affymetrix SNP array. A method for imputation adapted to inbred line panels was developed and will enable imputations of data issued from resequencing (cf WP3). We collaborated to the development of a mid-density 16k SNP array that was used to genotyped newly DH lines from T4.2.

Task 4.4 Tools and methods for GWAS and GS – Leader : L. Moreau (INRA UMR GQE)
GWAS and GS raise new issues in terms of methods and tools especially for hybrid panels and structured populations. The task objectives are to contribute to the development of GWAS and GS methods and tools.
A GWAS pipeline was implemented. A specific work was done to address GWAS implementation in hybrid panels. It led to the development a new R-package (Laporte et al. Biometrics 2017) for kinship estimation and efficient algorithms for mixed-models including several variance components. For GS ongoing work focusses on the impact of population structure on GS efficiency and development of GS methods adapted to admixed populations. Work is currently done on the integration of GxE interactions in GS and GWAS.