Imputation of canine genotype array data using 365 whole-genome sequences improves power of genome-wide association studies.

Jessica J Hayward,Margret L Casal,Michael Boyle,Marta G Castelhano,Rory J Todhunter,Laura M Shannon,Kenneth W Simpson,Nathan B Sutter,Adam R Boyko,Vicki N Meyers-Wallen,Michelle E White,Sharon A Center,Gregory S Barsh

doi:10.1371/journal.pgen.1008003

Abstract

Genomic resources for the domestic dog have improved with the widespread adoption of a 173k SNP array platform and updated reference genome. SNP arrays of this density are sufficient for detecting genetic associations within breeds but are underpowered for finding associations across multiple breeds or in mixed-breed dogs, where linkage disequilibrium rapidly decays between markers, even though such studies would hold particular promise for mapping complex diseases and traits. Here we introduce an imputation reference panel, consisting of 365 diverse, whole-genome sequenced dogs and wolves, which increases the number of markers that can be queried in genome-wide association studies approximately 130-fold. Using previously genotyped dogs, we show the utility of this reference panel in identifying potentially novel associations, including a locus on CFA20 significantly associated with cranial cruciate ligament disease, and fine-mapping for canine body size and blood phenotypes, even when causal loci are not in strong linkage disequilibrium with any single array marker. This reference panel resource will improve future genome-wide association studies for canine complex diseases and other phenotypes.

Highlights

The modern domestic dog (Canis lupus familiaris) consists of over 500 breeds selected for diverse roles and subject to wildly different disease prevalences [1]
An increase in density can be achieved by the following: adding more SNPs to the CanineHD array, using whole genome sequencing (WGS), or using imputation to predict genotypes through the use of a reference panel created from WGS data
Imputation has been used in a canine within-breed genome-wide association studies (GWAS) of primary hypoadrenocorticism in the Standard poodle, resulting in an approximately 20-fold increase in SNP number, this did not lead to the identification of a significant association [13]

Summary

Introduction

The modern domestic dog (Canis lupus familiaris) consists of over 500 breeds selected for diverse roles and subject to wildly different disease prevalences [1]. A high quality reference genome [2,3,4] and affordable SNP genotyping arrays [5] have helped make the dog a powerful animal model for studying the genetics of complex traits and diseases. With an average spacing of 1 SNP every 13kb, the CanineHD array (Illumina, San Diego, CA) has been successfully implemented in many genome-wide association studies (GWAS), especially within single breeds where linkage disequilibrium (LD) often extends beyond 1Mb [for example, see [6,7]. Imputation has been used in a canine within-breed GWAS of primary hypoadrenocorticism in the Standard poodle, resulting in an approximately 20-fold increase in SNP number, this did not lead to the identification of a significant association [13]

Objectives

Methods

Results

Discussion

Conclusion