约束标准化线性回归法估计合成品种动物基因组品种构成

doi:10.3864/j.issn.0578-1752.2020.01.018

摘要/Abstract

摘要：

【背景】合成品种是由至少两种纯种（祖先）培育的新品种,旨在兼顾祖先品种的有利遗传特征,并且可以长期保持后代的杂种优势而不需要每个世代都杂交。合成品种的遗传稳定,不同于杂交群体,因而可以像纯种一样繁育。实践中,估计合成品种的祖先品种对每个动物个体基因组的遗传贡献比例,即基因组品种构成（genomic breeding composition, GBC）,在畜禽品种登记、品种培育历史和品种构成分析、品种保护和杂交优势预测等方面有着非常重要的意义。利用基因组SNP基因型数据,采用合适的数学模型和统计方法,可以鉴定现有纯种品种的动物个体或纯种品种在杂交个体基因组的遗传贡献比例,而估计合成品种GBC的方法和研究都较少。【目的】线性回归是估计GBC的常用方法之一,但也存在诸多的问题。本研究旨在提出和评估一种约束的标准化线性回归方法（restricted standardized linear regression, RSLR）,作为传统线性回归方法的改进方法,应用于估计合成品种动物个体的GBC。【方法】采用肉牛王牛（Beefmaster）及其3个祖先品种（婆罗门牛、海福特牛和短角牛）的GGP 50K SNP芯片所测定的基因型数据,通过计算其基因频率和欧氏距离,利用层次聚类分析方法解析了4个动物群体的遗传关系,然后提出了RSLR方法,估计合成品种动物个体GBC的原理和方法。为了检验该方法的估计效果,从基因型数据中选择了均匀分布的分别包含1 000、5 000、10 000、20 000、30 000、40 000个SNP以及3个祖先品种共有的47 900个SNP的7个子集,分别采用RSLR和传统线性回归（linear regression, LR）两种方法估计了4 323头肉牛王牛的GBC,并比较了两种方法的计算结果。【结果】聚类分析的结果与4个品种间的遗传关系相吻合,表明肉牛王牛与婆罗门牛的遗传关系最近,遗传距离小于其与海福特牛和短角牛的遗传距离。LR方法估计的GBC会低估婆罗门牛（0.459—0.462）和短角牛（0.208—0.212）对于肉牛王牛的基因组贡献,同时高估海福特牛（0.326—0.333）的基因组贡献。但RSLR方法估计的肉牛王牛GBC的平均值与3个祖先品种预期的基因组贡献比例比较吻合：婆罗门牛为0.497—0.503,海福特牛为0.262—0.274,短角牛为0.229—0.231。此外,LR方法估计GBC的标准差和变异系数明显大于用RSLR估计的结果。当SNP子集数量在20 000以上时,LR方法估计牛肉王牛的3个祖先品种婆罗门牛、海福特牛和短角牛基因组贡献的标准差分别为0.048、0.032和0.051—0.052,变异系数分别为10.46%—10.50%、9.61%—9.76%和23.94%—25.00%,而RSLR方法估计的标准差,3个祖先品种对应为0.021、0.021—0.022和0.024—0.025,变异系数分别为4.18%—4.20%、7.89%—8.33%以及10.26%—10.68%。【结论】用RSLR方法估计的合成品种肉牛王牛动物个体的GBC,比LR方法的估计结果更加准确,估计的结果比LR方法估计的结果更稳定,且估计的一致性也更好,可以作为线性回归方法的改进,应用于估计合成品种动物个体GBC。

关键词: SNP芯片, 线性回归, 合成品种, 基因组品种构成

Abstract:

【Background】A composite breed is made up of two or more purebreds (ancestries), designed to combine advantageous genetic characteristics from the ancestry breeds and to retain heterosis in future generations without crossbreeding. Unlike crossbred populations, composite variety can be maintained as a purebred. In practice, knowing the ratio of genomic contribution of an ancestry breed to individual composite animals, referred to as the genomic breed composition (GBC), is of importance in animal breed registration, tracing breeding history and population structure, breed conservation, and the prediction of heterosis. Using a set of genomic SNP genotype and an appropriate statistical model, GBC of a purebred or crossbred animal can be estimated. So far, studies on statistical methods devote to the estimation of GBC in composite breed are limited. Linear regression (LR) analysis was commonly used to estimated GBC of individual animals, but it had some limitations such as the coefficients of ancestral breeds does not add to 1.【Objective】The purpose of the present study was to propose and evaluate the use of restricted standardized regression analysis, as an improved approach of linear regression analysis to estimate GBC in composite animals. 【Method】The dataset consisted of 4 323 Beefmaster cattle and purebred animals belonging to their ancestry breeds, namely Brahman, Hereford and Shorthorn. All these animals were genotyped by GeneSeek Genomic Profiling (GGP) bovine 50K SNP chips. Allelic frequencies of each SNP and the Euclidean distance between breeds were computed for the four animal populations, and their genetic relationships were revealed by Hierarchical Clustering based on Euclidean distance of SNP allele frequencies among the four populations. Genomic breed composition of the 4 323 Beefmaster cattle were estimated using RSLR and LR, respectively, based on 7 SNP panels(1K, 5K, 10K, 20K, 30K, 40K, and all the common 47 900 SNP). 【Result】The results of the clustering analysis agreed well with the genetic relationships of Beefmaster and the three ancestral breeds, showing that Beefmaster was more related to Brahman than Herdford and Shorhorn. Linear regression analysis underestimated the genomic contribution ratios of Brahman cattle (0.459-0.462) and shorthorn cattle (0.208-0.212) and at the same time overestimated that of Hereford cattle (0.326-0.333) to Beefmaster cattle. In contrast, estimated GBC of the 4 323 Beefmaster cattle obtained by using RSLR agreed well with expected genomic contribution ratios of the three ancestry breeds, which were 0.497-0.503 for Brahman, 0.262-0.274 for Hereford, and 0.229-0.231 for Shorthorn, respectively. Furthermore, the standard deviations (SD) and coefficients of variance (CV) of GBC obtained by using LR were larger than those obtained using RSLR. With 20K or more SNPs as the reference panels, the SD of GBC estimated by using LR were 0.048 (Brahman), 0.032 (Hereford) and 0.051-0.052 (Shorthorn), and the corresponding CV were 10.46%-10.50% (Brahman), 9.61%-9.76% (Hereford) and 23.94%-25.00% (Shorthorn), respectively. Using RSLR, on the other hand, the SD of GBC pertaining to each of the three ancestry breeds were 0.021 (Brahman), 0.021-0.022(Hereford) and 0.024-0.025 (Shorthorn), and the responding CV were 4.18%-4.20% (Brahman), 7.89%-8.33% (Hereford) and 10.26%-10.68% (Shorthorn), correspondingly. 【Conclusion】The RSLR method provided more accurate and consistent estimates of GBC in the 4 323 Beefmaster cattle than the LR approach. It thus provided a new statistical method for the estimation of GBC in composite animals.

Key words: SNP chip, linear regression, composite breeds, genomic breed composition

何俊,李智,吴晓林. 约束标准化线性回归法估计合成品种动物基因组品种构成[J]. 中国农业科学, 2020, 53(1): 191-200.

Jun HE,Zhi LI,XiaoLin WU. Using Restricted Standardized Linear Regression Model to Estimate Genomic Breed Composition in Composite Breed Animals[J]. Scientia Agricultura Sinica, 2020, 53(1): 191-200.

0
/ / 推荐

导出引用管理器 EndNote|Reference Manager|ProCite|BibTeX|RefWorks

链接本文: https://www.chinaagrisci.com/CN/10.3864/j.issn.0578-1752.2020.01.018

https://www.chinaagrisci.com/CN/Y2020/V53/I1/191

图/表 5

表1

图1

表2

表3

图2

参考文献 31

[1]	刘文忠 . 家畜合成群体保留杂种优势的预测与培育效果评价. 遗传, 2009,31(8):791-798.
	LIU W Z . Prediction of retained heterosis and evaluation on breeding effects of composite livestock populations. Hereditas(Beijing), 2009,31(8):791-798.(in Chinese)
[2]	MARSHALL B H, BRIGGS D M . Modern Breeds of Livestock. 4th ed. New York: MacMillian Company, 1980.
[3]	何俊, 钱长嵩, RICHARD G Tait Jr, Stewart Bauck, 吴晓林 . SNP芯片数据估计动物个体基因组品种构成的方法及应用. 遗传, 2018,40(4):305-314.
	HE J, QIAN C S TAIT Jr R G, BAUCK S, WU X L . Estimating genomic breed composition of individual animals using selected SNPs. Hereditas (Beijing), 2018,40(4):305-314. (in Chinese)
[4]	WU X L, LIU R Z, SHI Q S, LIU X C, LI X, WU M S . Marker-assisted mating applied in in-situ conservation of indigenous animals in small populations: (1) Choosing mating schemes for maximum heterozygosity. Asian-Australian Journal of Animal Science, 2000,13(4):431-434.
[5]	杨子博, 王安邦, 冷苏凤, 顾正中, 周羊梅 . 小麦新品种淮麦33的遗传构成分析. 中国农业科学, 2018,51(17):3237-3248.
	YANG Z B, WANG A B, LENG S F GU Z Z, ZHOU Y M . Genetic analysis of the novel high-yielding wheat cultivar Huaimai33. Scientia Agricultura Sinica, 2018,51(17):3237-3248.( in Chinese)
[6]	VANRADEN P M, COOPER T A . Genomic evaluations and breed composition for crossbred U.S. dairy cattle. Interbull Bulletin. Orlando,Florida, 2015.
[7]	PRITCHARD J K, STEPHENS M, DONNELLY P . Inference of population structure using multilocus genotype data. Genetics, 2000,155(2):945-959.
[8]	HE J, GUO Y G, XU J, LI H, FULLER A, TAIT R G, WU X L, BAUCK S . Comparing SNP panels and statistical methods for estimating genomic breed composition of individual animals in ten cattle breeds. BMC Genetics, 2018, 19:56.
[9]	GOBENA M, ELZO M A, MATEESCU R G . Population structure and genomic breed composition in an Angus-Brahman crossbred cattle population. Frontier Genetics, 2018,9:90.
[10]	CHIANG C W K, GAJDOS Z K Z, KORN J M, KURUVILLA F G, BUTLER J L, HACKETT R, GUIDUCCI C, NGUYEN T T, WILKS R, FORRESTER T, HAIMAN C A, HENDERSON K D, LE MARCHAND L, HENDERSON B E, PALMERT M R, MCKENZIE C A, LYON H N, COOPER R S, ZHU X F, HIRSCHHORN J N . Rapid assessment of genetic ancestry in populations of unknown origin by genome-wide genotyping of pooled samples. PLoS Genetics, 2010,6(3):e1000866.
[11]	KUEHN L A, KEELE J W, BENNETT G L, MCDANELD T G, SMITH T P L, SNELLING W M, SONSTEGARD T S, THALLMAN R M . Predicting breed composition using breed frequencies of 50,000 markers from the US Meat Animal Research Center 2,000 Bull Project. Journal of Animal Science, 2011,89(6):1742-1750.
[12]	HULSEGGE B, CALUS M P, WINDIG J J, HOVING-BOLINK A H, MAURICE-VAN EIJNDHOVEN M H, HIEMSTRA S J . Selection of SNP from 50K and 777K arrays to predict breed of origin in cattle. Journal of Animal Science, 2013,91:5128-5134.
[13]	AKANNO E C, CHEN L, ABO-ISMAIL M K, CROWLEY J J, WANG Z, LI C, BASARAB J A, MACNEIL M D, PLASTOW G . Genomic prediction of breed composition and heterosis effects in Angus, Charolais, and Hereford crosses using 50K genotypes. Canadian Journal of Animal Science, 2017,97(3):431-438.
[14]	MCVEAN G . A Genealogical Interpretation Of Principal Components Analysis. PLoS Genetics. 2009,5(10):e1000686.
[15]	MA J, AMOS C I . Principal components analysis of population admixture. PLoS ONE, 2012,7(7):e40115.
[16]	LEWIS J, ABAS Z, DADOUSIS C, LYKIDIS D, PASCHOU P, DRINEAS P . Tracing cattle breeds with principal components analysis ancestry informative SNPs. PLoS ONE. 2011,6(4):e18007.
[17]	BANSAL V, LIBIGER O . Fast individual ancestry inference from DNA sequence data leveraging allele frequencies for multiple populations. BMC Bioinformatics, 2015,16:4.
[18]	ALEXANDER D H, LANGE K . Enhancements to the ADMIXTURE algorithm for individual ancestry estima-tion. BMC Bioinform, 2011,12:246.
[19]	DODDS K G, AUVRAY B, NEWMAN N S A, MCEWAN C J . Genomic breed prediction in New Zealand sheep. BMC Genetics, 2014,15:92
[20]	FUNKHOUSER S A, BATES R O, ERNST C W, NEWCOM D, STEIBEL J P . Estimation of genome-wide and locus-specific breed composition in pigs. Translational Animal Science, 2017,1(1):36-44.
[21]	GOBENA M, ELZO M A, MATEESCU R G . Population structure and genomic breed composition in an Angus-Brahman crossbred cattle population. Frontiers in Genetics, 2018,9:90.
[22]	SARGOLZAEI M, CHESNAIS J P, SCHENKEL F S . A new approach for efficient genotype imputation using information from relatives. BMC Genom, 2014,15:478.
[23]	HE J, GUO YG, XU JQ, LI H, FULLER A, RICHARD G JR, WU XL, BAUCK S . Estimating genomic breed composition of individual animals in ten cattle breeds: Comparison of SNP panels and statistical methodology//Proceedings of the 11th World Congress on Genetics Applied to Live-stock Production. New Zealand: Auckland, 2018, 684-687.
[24]	MURTAGH F, LEGENDRE P . Ward's hierarchical agglomerative clustering method: which algorithms implement Ward's criterion? Journal of Classification, 2014,31(3):274-295.
[25]	LEGENDRE P, LEGENDRE L . Numerical Ecology. 3rd ed. Developments in environmental modelling. 2012,24.
[26]	桑世飞, 王会, 梅德圣, 刘佳, 付丽, 王军, 汪文祥, 胡琼 . 利用全基因组SNP芯片分析油菜遗传距离与杂种优势的关系. 中国农业科学, 2015,48(12):2469-2478.
	SANG S F, WANG H, MEI D S, LIU J, FU L, WANG J, WANG W X, HU Q . Correlation analysis between heterosis and genetic distance evaluated by genome-wide SNP chip in Brassica napus. Scientia Agricultura Sinica, 2015,48(12):2469-2478.
[27]	WRIGHT S . Correlation and causation. Journal of Agricultural Research, 1921,20(7):557-585.
[28]	HAN T S, KOBAYASHI K . Mathematics of Information and Coding. Boston, MA,USA: American Mathematical Society, 2001.
[29]	WU X L, XU J Q, FENG G F, WIGGANS G R, TAYLOR J F, HE J, QIAN C S, QIU J S, SIMPSON B, WALKER J, BAUCK S . Optimal design of low-density SNP arrays for genomic prediction: algorithm and applications. PLoS ONE, 2016,11(9):e0161719.
[30]	ABDI H, WILLIAMS L J . Principal component analysis. Wiley Interdisciplinary Reviews Computational Statistics, 2010,2(4):433-459.
[31]	WU X L, XU J Q, FENG G F, WIGGANS G R, TAYLOR J F, HE J, QIAN C S, QIU J S, SIMPSON B, WALKER J, BAUCK S . Optimal design of low-density SNP arrays for genomic prediction: algorithm and applications. PLoS ONE, 2016,11(9):e0161719.

	群体 Population	动物数量 Number of animals	SNP数量 Number of SNPs
肉牛王牛 Beefmaster	4323	49463	0.310±0.141
婆罗门牛 Brahman	68	49463	0.248±0.139
短角牛 Shorthorn	1232	49463	0.282±0.145
海福特牛Hereford	2423	49463	0.270±0.150

染色体 Chromosome	SNP子集 SNP panel
染色体 Chromosome	500	1000	3000	5000	10000	20000	30000	40000	47900
0	35	69	205	341	682	1364	2045	2727	3265
1	26	53	161	268	536	1072	1608	2144	2567
2	24	47	140	235	469	938	1407	1876	2247
3	22	44	133	221	443	886	1330	1772	2123
4	20	40	119	198	396	791	1186	1582	1894
5	22	45	135	225	450	900	1351	1801	2157
6	22	43	129	215	429	859	1288	1718	2057
7	19	39	116	193	386	772	1157	1543	1848
8	19	37	113	189	379	757	1136	1514	1812
9	19	38	113	187	374	748	1123	1497	1793
10	18	36	109	182	364	728	1091	1455	1742
11	18	37	111	186	372	745	1118	1490	1785
12	15	30	88	146	291	581	871	1162	1391
13	16	31	95	159	318	636	954	1272	1523
14	15	31	92	154	308	616	925	1233	1477
15	16	31	93	154	309	618	926	1235	1478
16	13	27	80	133	266	533	800	1067	1279
17	13	26	78	130	260	520	779	1039	1244
18	13	26	80	133	265	530	796	1061	1271
19	13	25	75	126	252	504	755	1007	1205
20	14	28	84	139	279	557	836	1114	1335
21	12	24	72	121	242	484	727	969	1160
22	11	22	64	106	211	423	633	845	1011
23	10	20	62	104	208	416	625	832	998
24	11	23	67	111	223	446	668	892	1067
25	8	15	45	76	152	303	455	606	726
26	9	18	55	91	182	365	547	730	874
27	7	15	45	76	151	301	453	603	722
28	9	17	50	82	164	329	492	657	787
29	8	17	53	89	179	357	537	716	857
X	23	46	138	230	460	921	1381	1841	2205

方法 Method	SNP子集 SNP panel	婆罗门牛 Brahman				海福特牛 Hereford				短角牛 Shorthorn
方法 Method	SNP子集 SNP panel	平均数 Mean	标准差 SD	中位数 Median	变异系数CV(%)	平均数 Mean	标准差 SD	中位数 Median	变异系数CV(%)	平均数 Mean	标准差 SD	中位数 Median	变异系数CV(%)
LR	1000	0.462	0.060	0.465	12.99	0.326	0.054	0.327	16.56	0.212	0.073	0.206	34.43
	5000	0.462	0.050	0.465	10.82	0.322	0.037	0.323	11.49	0.216	0.055	0.210	25.46
	10000	0.463	0.049	0.466	10.58	0.322	0.034	0.322	10.56	0.215	0.053	0.206	24.65
	20000	0.459	0.048	0.463	10.46	0.328	0.032	0.328	9.76	0.213	0.051	0.206	23.94
	30000	0.457	0.048	0.461	10.50	0.330	0.032	0.331	9.70	0.213	0.052	0.204	24.41
	40000	0.459	0.048	0.463	10.46	0.333	0.032	0.333	9.61	0.208	0.051	0.200	24.52
	47900	0.459	0.048	0.464	10.46	0.333	0.032	0.333	9.61	0.208	0.052	0.199	25.00
RSLR	1000	0.497	0.029	0.498	5.84	0.274	0.036	0.275	13.14	0.229	0.038	0.227	16.59
	5000	0.501	0.023	0.501	4.59	0.267	0.025	0.267	9.36	0.231	0.027	0.229	11.69
	10000	0.503	0.022	0.504	4.37	0.262	0.023	0.261	8.78	0.235	0.025	0.233	10.64
	20000	0.502	0.021	0.502	4.18	0.264	0.022	0.265	8.33	0.234	0.025	0.231	10.68
	30000	0.500	0.021	0.501	4.20	0.266	0.022	0.267	8.27	0.234	0.024	0.231	10.26
	40000	0.501	0.021	0.501	4.19	0.266	0.022	0.267	8.27	0.233	0.024	0.230	10.30
	47900	0.501	0.021	0.501	4.19	0.266	0.021	0.266	7.89	0.234	0.024	0.231	10.26