模擬探討基因體選種在水稻育種計畫中之運用

Genomic Selection in Rice Breeding Programs-a Simulation Study

潘芃諭(Peng-Yu Pan) ； 黃永芬(Yung-Fen Huang)

《作物、環境與生物資訊》 16卷4期 (2019/12) Pp. 173-195

https://doi.org/10.30061/CEB.201912_16(4).0001

摘要

基因體選種為隨著分子標誌高通量化而成形之分子標誌輔助選種方法，利用覆蓋全基因體的大量分子標誌估計個體育種價，並依此選拔優良個體。本研究目的在於藉由電腦模擬探討基因體選種在水稻育種計畫中之運用，分別探究雙親本雜交及輪迴選種兩種情境。我們使用327個秈稻（Oryza sativa L. ssp. indica）優良品系之公開數據，包括產量、抽穗期及株高，與5,264個單核苷酸多型性分子標誌。首先以十折交叉驗證比較8個統計方法的預測準確度，結果顯示產量和株高最適用之模型為RR－BLUP，抽穗期則以BayesB最佳。接著，我們自327品系中選出5個產量表現優異之品系作為雙親本雜交之親本，在典型外表型選種架構下，於F_2及F_6世代以基因體選種進行產量選拔。模擬結果顯示，基因體選種較外表型選種以較高頻率選得性狀表現較優良之品系：針對產量、抽穗期、或是株高，10個雜交組合中分別有9、6、8個雜交組合中獲得產量較高、抽穗較早、或株高較矮之品系。輪迴選種以327品系中產量與抽穗期各前4名共8個品系作為親本，旨在增加產量及縮短抽穗期。結果顯示，隨著循環數增加，族群的產量與抽穗期皆有顯著改進，且對偶基因頻度於早期循環即大量固定，顯示基因體選種搭配輪迴選種能有效改良族群並快速累積有利對偶基因頻度。本研究結果顯示，若欲應用基因體選種於水稻育種，可先以輪迴選種改良族群、從中選出優良親本進行雜交，於F_2及F_6以既節省人力又精確的基因體選種選拔優良品系。

關鍵字

基因體選種 ； 雙親本雜交 ； 輪迴選種 ； 水稻（Oryza sativa）

並列摘要

Genomic selection (GS) is one of the methods of marker-assisted selection, using genome-wide markers to estimate the genomic estimated breeding value (GEBV) for each individual, based on which the selection is made. This study aimed to explore the applications of GS in a rice breeding program using computer simulation. Both bi-parental crosses and recurrent selection schemes were investigated. Publicly available data were collected from 327 indica rice (Oryza sativa L. ssp. indica), including yield, flowering time (FL) and plant height (PH), as well as 5,264 single nucleotide polymorphism markers. We first assayed the prediction accuracy of eight statistical models based on 10-fold cross-validation in order to select the most suitable prediction model for each trait. The results showed that RR-BLUP was the most suitable model for yield and PH while BayesB for FL. For the classic bi-parental phenotypic selection (PS) scheme, five lines with excellent yield were selected as crossing parents from the initial 327 lines and GS was applied at F_2 and F_6 populations. Compared with PS, individuals selected using GS showed better performance in a higher frequency: among 10 bi-parental crosses, nine, six, and eight crosses produced lines with higher yield, earlier heading date, and shorter plant height. In recurrent selection scheme, the top four lines in yield and the earliest four liens in FL from the initial 327 varieties, were used as parents. The results showed that the yield and FL of the population were significantly improved with the increase of cycle number, and the allele frequency is almost fixed at early cycle. This indicates that GS can effectively improve the population and quickly accumulate favorable alleles. Based on the present study, a possible breeding scheme using GS could be to first improve the breeding population using recurrent selection. From the breeding population, we can either purify lines for variety release or select parents for bi-parental crosses and apply GS at F_2 and F_6 populations. This strategy could be a way to save manpower while precisely select candidate lines.

並列關鍵字

genomic selection ； bi-parental cross ； recurrent selection ； rice (Oryza sativa)