UK dogs data from: Genome-wide association studies for canine hip dysplasia in single and multiple populations – implications and potential novel risk loci
Cite this dataset
Friedrich, Juliane et al. (2021). UK dogs data from: Genome-wide association studies for canine hip dysplasia in single and multiple populations – implications and potential novel risk loci [Dataset]. Dryad. https://doi.org/10.5061/dryad.h44j0zpkf
Background: Association mapping studies of quantitative trait loci (QTL) for canine hip dysplasia (CHD) can contribute to the understanding of the genetic background of this common and debilitating disease and might contribute to its genetic improvement. The power of association studies for CHD is limited by relatively small sample numbers for CHD records within countries, suggesting potential benefits of joining data across countries. However, this is complicated due to the use of different scoring systems across countries. In this study, we incorporated routinely assessed CHD records and genotype data of German Shepherd dogs from two countries (UK and Sweden) to perform genome-wide association studies (GWAS) within populations using different variations of CHD phenotypes. As phenotypes, dogs were either classified into cases and controls based on the Fédération Cynologique Internationale (FCI) five-level grading of the worst hip or the FCI grade was treated as an ordinal trait. In a subsequent meta-analysis, we added publicly available data from a Finnish population and performed the GWAS across all populations. Genetic associations for the CHD phenotypes were evaluated in a linear mixed model using 62,089 SNPs.
Results: Multiple SNPs with genome-wide significant and suggestive associations were detected in single-population GWAS and the meta-analysis. Few of these SNPs overlapped between populations or between single-population GWAS and the meta-analysis, suggesting that many CHD-related QTL are population-specific. More significant or suggestive SNPs were identified when FCI grades were used as phenotypes in comparison to the case-control approach. MED13 (Chr 9) and PLEKHA7 (Chr 21) emerged as novel positional candidate genes associated with hip dysplasia.
Conclusions: Our findings confirm the complex genetic nature of hip dysplasia in dogs, with multiple loci associated with the trait, most of which are population-specific. Routinely assessed CHD information collected across countries provide an opportunity to increase sample sizes and statistical power for association studies. While the lack of standardisation of CHD assessment schemes across countries poses a challenge, we showed that conversion of traits can be utilised to overcome this obstacle.
DNA was extracted from saliva samples collected with Performagene PG-100 swabs (UK dogs). The dogs were genotyped using the Illumina CanineHD Whole-Genome Genotyping BeadChip featuring 172,115 SNPs. For quality control, filtering was imposed in GenomeStudio version 2.0 for sample call rate > 90%, SNP call rate > 98%, reproducibility (GTS) > 0.6 and low or confounded signal characterised by AB R mean (mean normalized intensity of the AB cluster) > 0.3. SNPs were also filtered using PLINK version 1.9 (27,28) to remove those with minor allele frequency (MAF) < 0.05 and significant deviations from Hardy-Weinberg equilibrium (HWE) (Bonferroni-corrected p-value of 0.05 = 4.5 x 10-7), resulting in 78,088 autosomal SNPs.
Canine hip dysplasia records for genotyped UK dogs were provided by the British Kennel Club (KC): the hip status of the dogs was screened by the BVA/KC scheme. In this scheme, determined by the severity of HD-related measurements from normal to severe, aggregated scores for bilateral joints (total hip score; HS) are given from 0 to 106 (0 to 53 for each joint). BVA/KC total hip scores for UK dogs were also converted into FCI five-level grades following a recommended conversion (0-10=A, 11-25=B, 26-35=C, 36-50=D, 51-106=E). As third phenotype approach, UK dogs were grouped into "cases" (BVA/KC scores >=11) and "controls" (scores 0-10).
Information on the data sets can be found in the "README.txt" file.