Animal & Grassland Research and Innovation Centre, Teagasc, Moorepark, Co. Cork, Ireland.
Animal. 2012 Aug;6(8):1206-15. doi: 10.1017/S1751731112000067.
Genome-wide association studies for difficult-to-measure traits are generally limited by the sample population size with accurate phenotypic data. The objective of this study was to utilise data on primiparous Holstein-Friesian cows from experimental farms in Ireland, the United Kingdom, the Netherlands and Sweden to identify genomic regions associated with traditional measures of fertility, as well as a fertility phenotype derived from milk progesterone profiles. Traditional fertility measures investigated were days to first heat, days to first service, pregnancy rate to first service, number of services and calving interval (CI); post-partum interval to the commencement of luteal activity (CLA) was derived using routine milk progesterone assays. Phenotypic and genotypic data on 37 590 single nucleotide polymorphisms (SNPs) were available for up to 1570 primiparous cows. Genetic parameters were estimated using linear animal models, and univariate and bivariate genome-wide association analyses were undertaken using Bayesian stochastic search variable selection performed using Gibbs sampling. Heritability estimates of the traditional fertility traits varied from 0.03 to 0.16; the heritability for CLA was 0.13. The posterior quantitative trait locus (QTL) probabilities, across the genome, for the traditional fertility measures were all <0.021. Posterior QTL probabilities of 0.060 and 0.045 were observed for CLA on SNPs each on chromosome 2 and chromosome 21, respectively, in the univariate analyses; these probabilities increased when CLA was included in the bivariate analyses with the traditional fertility traits. For example, in the bivariate analysis with CI, the posterior QTL probability of the two aforementioned SNPs were 0.662 and 0.123. Candidate genes in the vicinity of these SNPs are discussed. The results from this study suggest that the power of genome-wide association studies in cattle may be increased by sharing of data and also possibly by using physiological measures of the trait under investigation.
对于难以测量的性状，全基因组关联研究通常受到具有准确表型数据的样本群体大小的限制。本研究的目的是利用来自爱尔兰、英国、荷兰和瑞典实验农场的头胎荷斯坦-弗里生奶牛的数据，鉴定与传统生育力指标以及源自牛奶孕酮谱的生育力表型相关的基因组区域。研究中调查的传统生育力指标包括：首次发情天数、首次配种天数、受胎率、配种次数和产犊间隔（CI）；产后黄体活动（CLA）间隔使用常规牛奶孕酮测定法推导得出。可获得多达 1570 头初产奶牛的 37590 个单核苷酸多态性（SNP）的表型和基因型数据。使用线性动物模型估计遗传参数，并使用贝叶斯随机搜索变量选择进行单变量和双变量全基因组关联分析，使用 Gibbs 抽样进行。传统生育力性状的遗传力估计值在 0.03 到 0.16 之间；CLA 的遗传力为 0.13。整个基因组中，传统生育力指标的后验数量性状基因座（QTL）概率均<0.021。在单变量分析中，分别在第 2 号和第 21 号染色体上的 SNP 上观察到 CLA 的后验 QTL 概率为 0.060 和 0.045；当 CLA 包含在与传统生育力性状的双变量分析中时，这些概率增加。例如，在与 CI 的双变量分析中，上述两个 SNP 的后验 QTL 概率为 0.662 和 0.123。讨论了这些 SNP 附近的候选基因。本研究结果表明，通过共享数据并可能通过使用所研究性状的生理测量值，可能会增加牛的全基因组关联研究的效力。
