Genomic selection (GS) consists in predicting breeding values of selection candidates, using a large number of genetic markers. An important question in GS is to determine the number of markers required for a good prediction. For this purpose, we introduce new proxies for the accuracy of the prediction. These proxies are suitable under sparse genetic map where it is likely to observe some imperfect linkage disequilibrium, that is, the situation where the alleles at a gene location and at a marker located nearby vary. Moreover, our suggested proxies are helpful for designing cost-effective SNP chips based on a moderate density of markers. We analyse rice data from Los Banos, Philippines and focus on the flowering time collected during the dry season 2012. Using different densities of markers, we show that at least 1553 markers are required to implement GS. Finding the optimal number of markers is crucial in order to optimize the breeding program.

中文翻译：

---

不完全连锁不平衡下高维线性模型的预测及其在基因组选择中的应用

基因组选择 (GS) 包括使用大量遗传标记预测选择候选者的育种值。GS 中的一个重要问题是确定良好预测所需的标记数量。为此，我们为预测的准确性引入了新的代理。这些代理适用于可能观察到一些不完全连锁不平衡的稀疏遗传图谱下，即基因位置和位于附近的标记处的等位基因变化的情况。此外，我们建议的代理有助于基于中等密度的标记设计具有成本效益的 SNP 芯片。我们分析了菲律宾洛斯巴诺斯的水稻数据，重点关注 2012 年旱季收集的开花时间。使用不同密度的标记，我们表明至少需要 1553 个标记才能实现 GS。找到最佳标记数量对于优化育种计划至关重要。