Metabolomic prediction of yield in hybrid rice
Metabolomic prediction of yield in hybrid rice
复制标题
杂交水稻产量的代谢组学预测
DOI：
10.1111/tpj.13242
复制
发表时间：
2016-10-01
期刊：
影响因子：
7.2
通讯作者：
Zhang, Qifa
中科院分区：
文献类型：
--
作者：
Xu, Shizhong;Xu, Yang;Zhang, Qifa
水稻（Oryza sativa）是世界上50%以上人口的主食。提高产量可大大促进全球粮食安全。杂交育种可能有助于实现这一目标，因为与纯种品种相比，杂交水稻的产量通常会有相当大的增加。我们最近开发了一种标记指导的杂交种产量预测方法，并通过基因组杂交育种显示产量大幅增加。我们现在有转录组学和代谢组学数据作为预测的潜在资源。利用最小绝对收缩选择算子（LASSO）、最佳线性无偏预测（BLUP）、随机搜索变量选择、偏最小二乘法以及基于径向基函数和多项式核函数的支持向量机等6种预测方法，我们发现利用这些组学数据可以进一步提高杂交种产量的可预测性。LASSO和BLUP是产量预测最有效的方法。对于高遗传力性状，基因组数据仍然是最有效的预测因子。当使用代谢组学数据时，与基因组预测相比，杂交产量的可预测性几乎增加了一倍。从210个重组自交系衍生的21945个潜在的杂交种中，从代谢产物预测的前10个杂交种的选择将导致产量增加30%。我们假设，每种代谢物代表了一个生物学上内置的遗传网络的产量，因此，使用代谢物进行预测相当于使用信息集成从这些隐藏的遗传网络的产量预测。
Rice (Oryza sativa) provides a staple food source for more than 50% of the world's population. An increase in yield can significantly contribute to global food security. Hybrid breeding can potentially help to meet this goal because hybrid rice often shows a considerable increase in yield when compared with pure-bred cultivars. We recently developed a marker-guided prediction method for hybrid yield and showed a substantial increase in yield through genomic hybrid breeding. We now have transcriptomic and metabolomic data as potential resources for prediction. Using six prediction methods, including least absolute shrinkage and selection operator (LASSO), best linear unbiased prediction (BLUP), stochastic search variable selection, partial least squares, and support vector machines using the radial basis function and polynomial kernel function, we found that the predictability of hybrid yield can be further increased using these omic data. LASSO and BLUP are the most efficient methods for yield prediction. For high heritability traits, genomic data remain the most efficient predictors. When metabolomic data are used, the predictability of hybrid yield is almost doubled compared with genomic prediction. Of the 21 945 potential hybrids derived from 210 recombinant inbred lines, selection of the top 10 hybrids predicted from metabolites would lead to a similar to 30% increase in yield. We hypothesize that each metabolite represents a biologically built-in genetic network for yield; thus, using metabolites for prediction is equivalent to using information integrated from these hidden genetic networks for yield prediction.