全基因體預測用於水稻雙親本雜交的親本決定

Abstract

親本的選擇對於雙親本雜交育種來說是很重要的步驟，一組良好的親本能夠在育種成之後產生表現優良的重組自交系（recombinant inbred line, RIL）。在本研究中，我們提出了一個基於全基因組預測（genomic prediction, GP）的方法來進行親本的挑選，而全基因組預測同時也用於估計親本候選族群和重組自交系的基因組育種價（genomic estimated breeding value, GEBV）。透過高通量水稻基因組數據集的模擬研究來分析一些選擇親本的策略，其結果顯示，最佳策略必須是同時考慮GEBV和親本的基因型多樣性。在研究中，我們為21種目標性狀提供一組最佳的親本作為參考，此外，我們同時也研究了六組兩個目標性狀組合親本選擇的最佳策略。我們提出的系統方法能夠適用於其他自交作物，並且很容易拓展到三個或以上個目標性狀。本文中亦提供R程式供使用者執行分析及模擬的過程。

The determination of parental lines is the first and most important step to a successful bi-parental crossing plant breeding program. A set of superior parental lines can lead to high performing recombinant inbred lines (RILs). In this study, we propose to select parental lines of rice based on genomic prediction (GP). The GP is applied to predict genomic estimated breeding values (GEBVs) for all the candidate parental lines and the RILs after several generations of self-pollinating. Some strategies of selecting the parental lines are investigated through simulation studies based on a high-quality rice genome dataset. It is shown that the best strategy in general takes both the GEBVs and the genomic diversity of parental lines into account. In this study, we present a set of parental lines for each of 21 quantitative traits. We also investigate the best selection strategy for 6 different combinations of two target traits. Our proposed systematic analysis procedure can be applicable to other self-pollinated crops, and it is readily extended to the more complex multi-trait situations with three or more target traits. Some R functions are provided for users to exercise the analysis procedure.