RETRACTED ARTICLE: Dissecting closely linked association signals in combination with the mammalian phenotype database can identify candidate genes in dairy cattle

Abstract

BackgroundGenome-wide association studies (GWAS) have been successfully implemented in cattle research and breeding. However, moving from the associations to identifying the causal variants and revealing underlying mechanisms have proven complicated. In dairy cattle populations, we face a challenge due to long-range linkage disequilibrium (LD) arising from close familial relationships in the studied individuals. Long range LD makes it difficult to distinguish if one or multiple quantitative trait loci (QTL) are segregating in a genomic region showing association with a phenotype. We had two objectives in this study: 1) to distinguish between multiple QTL segregating in a genomic region, and 2) use of external information to prioritize candidate genes for a QTL along with the candidate variant.ResultsWe observed fixing the lead SNP as a covariate can help to distinguish additional close association signal(s). Thereafter, using the mammalian phenotype database, we successfully found candidate genes, in concordance with previous studies, demonstrating the power of this strategy. Secondly, we used variant annotation information to search for causative variants in our candidate genes. The variant information successfully identified known causal mutations and showed the potential to pinpoint the causative mutation(s) which are located in coding regions.ConclusionsOur approach can distinguish multiple QTL segregating on the same chromosome in a single analysis without manual input. Moreover, utilizing information from the mammalian phenotype database and variant effect predictor as post-GWAS analysis could benefit in candidate genes and causative mutations finding in cattle. Our study not only identified additional candidate genes for milk traits, but also can serve as a routine method for GWAS in dairy cattle.Electronic supplementary materialThe online version of this article (10.1186/s12863-018-0620-0) contains supplementary material, which is available to authorized users.

Prioritizing candidate genes post-GWAS using multiple sources of data for mastitis resistance in dairy cattle

“…Using our previously described GWAS approach [ 27 ], we ran an association analysis with imputed whole-genome sequencing (WGS) variants for resistance to CM in Nordic Holstein cattle. Figure 1 presents the Manhattan plot of the GWAS results.…”

“…We found 22 independent association signals using imputed WGS variants and CM as a phenotype. Our association analysis approach, which was described previously [ 27 ], aids in dissecting closely linked QTLs. The 22 lead SNPs overlapped or were close to many previously reported QTLs [ 10 , 12 , 14 , 44 ] and explained 14% of the variance for de-regressed breeding value for resistance to CM.…”

“…The SNP-wise Mean ( snp-wise = mean) model is more attuned to the mean SNP association, but is skewed to high-LD regions. In our experience, the lead SNP is not always the causal SNP [ 27 ]. Thus, we wanted to avoid calculating the gene P-value based on the top SNP in genes ( snp-wise = top ), an option that uses only a fraction of the SNPs from each gene.…”

“…Details of the association analysis pipeline and simulation verification were reported previously [ 27 ]. In the first round, we used GCTA [ 16 ] to perform single-SNP GWAS analysis on each chromosome.…”

Retraction Note: Dissecting closely linked association signals in combination with the mammalian phenotype database can identify candidate genes in dairy cattle

“…This article [ 1 ] has been retracted by the authors after discovering that a set of markers which would not pass quality control criteria has been used for the association study described, resulting in false positives being included. As a result, some sections of the article are inaccurate.…”