Imputation-Based Fine-Mapping Suggests That Most QTL in an Outbred Chicken Advanced Intercross Body Weight Line Are Due to Multiple, Linked Loci

Abstract

The Virginia chicken lines have been divergently selected for juvenile body weight for more than 50 generations. Today, the high- and low-weight lines show a >12-fold difference for the selected trait, 56-d body weight. These lines provide unique opportunities to study the genetic architecture of long-term, single-trait selection. Previously, several quantitative trait loci (QTL) contributing to weight differences between the lines were mapped in an F2-cross between them, and these were later replicated and fine-mapped in a nine-generation advanced intercross of them. Here, we explore the possibility to further increase the fine-mapping resolution of these QTL via a pedigree-based imputation strategy that aims to better capture the genetic diversity in the divergently selected, but outbred, founder lines. The founders of the intercross were high-density genotyped, and then pedigree-based imputation was used to assign genotypes throughout the pedigree. Imputation increased the marker density 20-fold in the selected QTL, providing 6911 markers for the subsequent analysis. Both single-marker association and multi-marker backward-elimination analyses were used to explore regions associated with 56-d body weight. The approach revealed several statistically and population structure independent associations and increased the mapping resolution. Further, most QTL were also found to contain multiple independent associations to markers that were not fixed in the founder populations, implying a complex underlying architecture due to the combined effects of multiple, linked loci perhaps located on independent haplotypes that still segregate in the selected lines.

A complex multi-locus, multi-allelic genetic architecture underlying the long-term selection-response in the Virginia body weight line of chickens

“…Reported here is our most recent progress in dissection of the genetic architecture underlying the extreme adaptations in the Virginia body weight chicken lines. Seven major QTL, mapped and confirmed in earlier reports (Besnier et al 2011; Brandt et al 2016; Jacobsson et al 2005; Wahlberg et al 2009), were fine-mapped using data from the F 15 generation of an Advanced Intercross Line bred from HWS and LWS founders from generation 40. We performed a multi-locus, association-based fine-mapping analysis, and explored the changes in allele frequencies across the QTL during ongoing selection in the parental lines.…”

“…As highlighted in Churchill (Churchill 2016), efforts to map quantitative trait loci (QTL) have often revealed an almost overwhelming complexity. Fine-mapping efforts have been challenged by the highly polygenic architecture of traits, where each of the contributing loci often encompass additional complexities, including multiple tightly linked variants with small or large, additive or epistatic genetic effects (; Brandt et al 2016; Holland 2007; Laurie et al 2004; Mott and Flint 2013). Therefore, there is a shortage of studies that provide a deeper understanding about the genetic architectures contributing to the adaptive response to longterm selection (Burke 2012).…”

“…Modern genomics allows cost-efficient, in-depth characterization of genetic variation across many loci, or entire genomes, in large populations. This provides opportunities to design studies that make efficient use of selected populations to gain a genome-wide perspective on the genetic architecture of long-term selection response (Burke 2012; Johansson et al 2010; Pettersson et al 2013) or to dissect the complexity within and across the loci that contribute to adaptations for highly polygenic traits (Brandt et al 2016; Chan et al 2012; Sheng et al 2015). Here, we describe the simultaneous fine-mapping of nine QTL contributing to longterm selection responses in the Virginia chicken lines.…”

“…After 40 generations of selection, the High- (HWS) and Low- (LWS) weight selected lines differed 9-fold for the selected trait. In the current generation (S59) the lines differ by more than 16-fold with many loci contributing to this difference (Besnier et al 2011; Carlborg et al 2006; Jacobsson et al 2005; Johansson et al 2010; Pettersson et al 2011; Sheng et al 2015; Wahlberg et al 2009) with likely contributions by allelic heterogeneity (Besnier et al 2011; Brandt et al 2016) and epistasis (Pettersson et al 2011).…”

A multi-locus association analysis method integrating phenotype and expression data reveals multiple novel associations to flowering time variation in wild-collectedArabidopsis thaliana

“…We have earlier developed a multi locus approach to explore the genetic architecture of 326 highly polygenic traits and a detailed description of the approach is available in (Sheng et al 327 2015;Brandt et al 2017;Lillie et al 2017). In short, the method was designed to study traits 328 where earlier data suggest them to be polygenic in the analysed population.…”