Association analysis for udder health based on SNP-panel and sequence data in Danish Holsteins

Abstract

BackgroundThe sensitivity of genome-wide association studies for the detection of quantitative trait loci (QTL) depends on the density of markers examined and the statistical models used. This study compares the performance of three marker densities to refine six previously detected QTL regions for mastitis traits: 54 k markers of a medium-density SNP (single nucleotide polymorphism) chip (MD), imputed 777 k markers of a high-density SNP chip (HD), and imputed whole-genome sequencing data (SEQ). Each dataset contained data for 4496 Danish Holstein cattle. Comparisons were performed using a linear mixed model (LM) and a Bayesian variable selection model (BVS).ResultsAfter quality control, 587, 7825, and 78 856 SNPs in the six targeted regions remained for MD, HD, and SEQ data, respectively. In general, the association patterns between SNPs and traits were similar for the three marker densities when tested using the same statistical model. With the LM model, 120 (MD), 967 (HD), and 7209 (SEQ) SNPs were significantly associated with mastitis, whereas with the BVS model, 43 (MD), 131 (HD), and 1052 (SEQ) significant SNPs (Bayes factor > 3.2) were observed. A total of 26 (MD), 75 (HD), and 465 (SEQ) significant SNPs were identified by both models. In addition, one, 16, and 33 QTL peaks for MD, HD, and SEQ data were detected according to the QTL intensity profile of SNP bins by post-analysis of the BVS model.ConclusionsThe power to detect significant associations increased with increasing marker density. The BVS model resulted in clearer boundaries between linked QTL than the LM model. Using SEQ data, the six targeted regions were refined to 33 candidate QTL regions for udder health. The comparison between these candidate QTL regions and known genes suggested that NPFFR2, SLC4A4, DCK, LIFR, and EDN3 may be considered as candidate genes for mastitis susceptibility.Electronic supplementary materialThe online version of this article (doi:10.1186/s12711-015-0129-1) contains supplementary material, which is available to authorized users.

Genome-wide association study for longevity with whole-genome sequencing in 3 cattle breeds

“…META-CHR6-88MB overlaps with 2 genes: neuropeptide G-protein coupled receptor gene (NPFFR2; 89,052,210-89,059,348 bp) and vitamin D-binding protein precursor (GC;88,695,739,180 bp). This genomic region was reported to harbor mastitis QTL in HOL and RDC (Wu et al, 2015), suggesting that multiple causal variants for mastitis may be located in this region. The second most significantly associated region was META-CHR18-58MB, located very near the zinc finger protein 613 gene (ZNF613;58,115,117,110 bp) and zinc finger protein 717 gene (ZNF717; 58,130,465-58,141,877 bp), which have been associated with calving difficulties.…”

Genetic susceptibility to infectious diseases: Current status and future perspectives from genome-wide approaches

“…Efforts to identify susceptibility alleles for other infectious diseases detected small-effect loci with poor replicability (reviewed in (Raszek et al, 2016)). To date, the greatest success was obtained for mastitis, which is commonly caused by bacterial infections; regions spanning the DCK, SLC4A4, and EDN3 genes were detected in at least two studies (Kanazawa et al, 1989;Sahana et al, 2013;Sodeland et al, 2011;Wu et al, 2015). The functional role of these genes in mastitis and response to invading bacteria remains to be evaluated.…”

Milk progesterone measures to improve genomic selection for fertility in dairy cows

“…The RWAS was able to refine the QTL regions from the GWAS, but it was not possible to identify the causative mutation, mainly because of long-range LD that exist in cattle due to low effective population size and strong selection. Similar observations have been reported in previous studies Wu et al, 2015). Another factor that might be hampering identification of the causative mutation is that imputation is not 100% accurate, especially for rare variants and small training populations.…”

“…Therefore, imputation from lower density genotypes to whole genome sequence using a sequenced training population offers a good alternative. The approach that has been taken in chapter 3 was to perform a GWAS with a low density (e.g., here 85k) SNP chip panel, and then focus on the identified peaks, performing a region-wise association study (RWAS) using imputed sequence data (e.g., Sahana et al, 2014;Wu et al, 2015). In chapter 3, significant QTL regions from the GWAS with 85k SNP were fine-mapped for endocrine fertility traits using imputed sequence variants.…”