肉牛多品种全基因组选择关键技术

This project, considering the actual situation of Chinese beef cattle breeding, will mainly research on enlarging reference population, model selection, and the LD similarity between breeds, computing methods and small scale SNP chip with high efficiency. We will build a reference population, consisted of one purebred and three cross herds, with the total number of 2400 younger bulls, as to it can be used for multiply breed GEBV estimation. The GEBV accuracy and computing time will expected to be improved by constructing trait-specific haplotype, hierarchical design, cluster analysis, and reduce the model. The GBS (genotyping by sequencing) will be used for finding high density SNPs which are tightly linked to QTL, and then, the result was used to design high efficient DNA chip. These research conclusions will be integrated into an improved GS method and applied to construct Chinese beef cattle GS system.

本研究针对目前肉牛全基因组选择（GS）研究中遇到的纯种参考群体应用受限、GEBV的估计准确度差、运算速度慢等瓶颈问题，充分考虑我国肉牛育种的实际情况，扩建现有参考群体，重点进行模型优化、群体间LD相似程度研究、算法筛选、小规格芯片位点筛选、LD片段与单个标记GEBV对比等内容，提高混合品种参考群体的应用范围；通过性状特异标记的亲缘关系阵、特异标记片段(单倍型)亲缘关系阵的构建，利用分层设计、聚类分析等约化模型技术，提高GEBV的估计准确度和运算速度；应用基于测序的SNP基因型(GBS)检测技术增加标记密度，寻找与肉牛主要经济性状QTL紧密连锁的一批标记，并设计高效的小规模芯片，突破现有技术瓶颈，最终研制一套适合我国肉牛育种的GS技术，实现肉牛的GS多品种选择。

本项目完善了基因组选择的方法，提高了基因组育种值估计的准确性，选用了GBLUP、BayesA和BayesB三种算法对目标性状基因组育种值进行了估计，构建西门塔尔牛群体的标记效应图谱，并探讨标记密度对基因组估计准确性的影响。计算结果表明：BayesB方法在屠宰性状上具有最高的准确性。在GBLUP方法中，用于计算G阵的标记数目在50000-100000之间时基因组育种值具有较高的准确性。在多性状模型在基因组选择上的研究中，选用了mtGBLUP和mtBayseA方法对目标性状进行了育种值估计并与单性状进行了比较，结果表明多性状基因组选择的方法估计育种值较单性状有一定的提高。此外，我们对基因组选择中的BayesB和BayesCπ方法进进行了改进，使得准确性分别提高0.3%-1%，和0.04% ～0.17%。