比较机器学习等算法对肉鸡产蛋性状育种值估计的准确性

doi:10.11843/j.issn.0366-6964.2022.05.005

摘要/Abstract

摘要： 我国白羽肉鸡育种中，通过遗传途径提高产蛋数和控制合适的蛋重是培育优良品系的一个重要方面。为探索适合我国白羽肉鸡育种中的基因组选择模型，本研究以2 474只白羽肉鸡品系的产蛋性状为研究对象，主要分析了机器学习算法KAML、BLUP(包括：PBLUP、GBLUP、SSGBLUP)和Bayes(包括：Bayes A、Bayes B和Bayes Cπ)方法对产蛋数和蛋重性状的预测准确性，准确性以5倍交叉验证进行评估。利用系谱以及基因组信息估计了产蛋数和蛋重性状的遗传力和遗传相关。结果表明，产蛋数性状遗传力为0.061~0.16，属于低遗传力性状；蛋重遗传力为0.28~0.39，属于中等遗传力性状；产蛋数与蛋重是中等遗传负相关(-0.518~-0.184)，不同阶段产蛋数之间是强的遗传正相关(0.736~0.998)。不同模型预测43周产蛋数和52周蛋重的育种值估计准确性结果表明，KAML方法对两者的预测准确性分别为0.115和0.266，与GBLUP方法(准确性分别为0.118和0.283)和SSGBLUP方法(准确性分别为0.136和0.259)的准确性差异显著，同时显著低于Bayes方法(准确性分别为0.230~0.239、0.336~0.340)的预测准确性， PBLUP方法预测准确性最低(准确性分别为0.095和0.246)。因此，在白羽肉鸡产蛋数和蛋重性状中应用Bayes方法将获得最高的育种值估计准确性。

关键词: 白羽肉鸡, 产蛋性状, 基因组选择, 机器学习, 遗传参数

Abstract: It is an important aspect to raise the number of eggs and control the appropriate egg weight by genetic way in the breeding of white-feather broilers in China. In order to explore the genome selection model suitable for the breeding of Chinese white-feather broilers, this study took 2 474 white-feather broilers, mainly analyzed the prediction accuracy of the machine learning algorithms KAML, BLUP (including:PBLUP, GBLUP, SSGBLUP) and Bayes (Bayes A, Bayes B and Bayes Cπ) methods in predicting egg number and egg weight traits. And the prediction accuracy was evaluated by 5-fold cross validation. The heritability and genetic correlation of the traits of egg number and egg weight were estimated using pedigree and genomic information. The results showed that the heritability of egg number was 0.061-0.16, which was a low heritability trait. The heritability of egg weight was 0.28-0.39, which was a medium heritability trait. The egg number and egg weight had a medium negative genetic correlation (-0.518——0.184). There was a strong positive genetic correlation (0.736-0.998) between the number of eggs laid at different stages. The accuracy of breeding value estimation for egg number (43 weeks) and egg weight (52 weeks)predicted by different models showed that the prediction accuracy of KAML method were 0.115 and 0.266, respectively, which were significantly different from that of GBLUP method (0.118 and 0.283) and SSGBLUP method (0.136 and 0.259), and were significantly lower than that of Bayes method (accuracy:0.230-0.239, 0.336-0.340). PBLUP method had the lowest prediction accuracy (0.095 and 0.246, respectively). Therefore, the Bayes method for the egg number and egg weight traits of white-feather broilers will achieve the highest accuracy of breeding value estimation.

Key words: white-feather broilers, laying trait, genomic selection, machine learning, genetic parameter

中图分类号:

S831.2

丁纪强, 李庆贺, 张高猛, 李森, 郑麦青, 文杰, 赵桂苹. 比较机器学习等算法对肉鸡产蛋性状育种值估计的准确性[J]. 畜牧兽医学报, 2022, 53(5): 1364-1372.

DING Jiqiang, LI Qinghe, ZHANG Gaomeng, LI Sen, ZHENG Maiqing, WEN Jie, ZHAO Guiping. Comparing the Accuracy of Estimated Breeding Value by Several Algorithms on Laying Traits in Broilers[J]. Acta Veterinaria et Zootechnica Sinica, 2022, 53(5): 1364-1372.

参考文献 46

[1]	HENDERSON C R.Best linear unbiased estimation and prediction under a selection model[J].Biometrics,1975,31(2):423-447.
[2]	MEUWISSEN T H E,HAYES B J,GODDARD M E.Prediction of total genetic value using genome-wide dense marker maps[J].Genetics,2001,157(4):1819-1829.
[3]	赵志达,张莉.基因组选择在绵羊育种中的应用[J].遗传,2019,41(4):293-303.ZHAO Z D,ZHANG L.Applications of genome selection in sheep breeding[J].Hereditas (Beijing),2019,41(4):293-303.(in Chinese)
[4]	WIGGANS G R,COLE J B,HUBBARD S M,et al.Genomic selection in dairy cattle:the USDA experience[J].Annu Rev Anim Biosci,2017,5:309-327.
[5]	WOLC A,STRICKER C,ARANGO J,et al.Breeding value prediction for production traits in layer chickens using pedigree or genomic relationships in a reduced animal model[J].Genet Sel Evol,2011,43(1):5.
[6]	LIU T F,QU H,LUO C L,et al.Genomic selection for the improvement of antibody response to Newcastle disease and avian influenza virus in chickens[J].PLoS One,2014,9(11):e112685.
[7]	WOLC A,DROBIK-CZWARNO W,FULTON J E,et al.Genomic prediction of avian influenza infection outcome in layer chickens[J].Genet Sel Evol,2018,50(1):21.
[8]	YIN L L,ZHANG H H,ZHOU X,et al.KAML:improving genomic prediction accuracy of complex traits using machine learning determined parameters[J].Genome Biol,2020,21(1):146.
[9]	LIU R R,XING S Y,WANG J,et al.A new chicken 55K SNP genotyping array[J].BMC Genomics,2019,20(1):410.
[10]	PURCELL S,NEALE B,TODD-BROWN K,et al.PLINK:a tool set for whole-genome association and population-based linkage analyses[J].Am J Hum Genet,2007,81(3):559-575.
[11]	BROWNING S R,BROWNING B L.Rapid and accurate haplotype phasing and missing-data inference for whole-genome association studies by use of localized haplotype clustering[J].Am J Hum Genet,2007,81(5):1084-1097.
[12]	VANRADEN P M.Efficient methods to compute genomic predictions[J].J Dairy Sci,2008,91(11):4414-4423.
[13]	MISZTAL I,AGUILAR I,LEGARRA A,et al.Choice of parameters for single-step genomic evaluation for type[J].J Dairy Sci,2010,93(S1):166.
[14]	GILMOUR A R,THOMPSON R,CULLIS B R.Average information REML:an efficient algorithm for variance parameter estimation in linear mixed models[J].Biometrics,1995,51(4):1440-1450.
[15]	梁小军,高慧兰,马吉锋,等.中国荷斯坦奶牛月龄和胎间距的回归分析[J].黑龙江畜牧兽医,2018(10):92-95,240.LIANG X J,GAO H L,MA J F,et al.Regression analysis of monthly age and spacing of Chinese Holstein dairy cows[J].Heilongjiang Animal Science and Veterinary Medicine,2018(10):92-95,240. (in Chinese)
[16]	BUZAŁA M,JANICKI B,CZARNECKI R.Consequences of different growth rates in broiler breeder and layer hens on embryogenesis,metabolism and metabolic rate:a review[J].Poult Sci,2015,94(4):728-733.
[17]	LIU Z,SUN C J,YAN Y Y,et al.Genome-wide association analysis of age-dependent egg weights in chickens[J].Front Genet,2018,9:128.
[18]	LIU Z,YANG N,YAN Y Y,et al.Genome-wide association analysis of egg production performance in chickens across the whole laying period[J].BMC Genetics,2019,20(1):67.
[19]	NIKNAFS S,NEJATI-JAVAREMI A,MEHRABANI-YEGANEH H,et al.Estimation of genetic parameters for body weight and egg production traits in Mazandaran native chicken[J].Trop Anim Health Prod,2012,44(7):1437-1443.
[20]	LI W,ZHENG M Q,ZHAO G P,et al.Identification of QTL regions and candidate genes for growth and feed efficiency in broilers[J].Genet Sel Evol,2021,53(1):13.
[21]	ALDRIDGE M N,VANDENPLAS J,BERGSMA R,et al.Variance estimates are similar using pedigree or genomic relationships with or without the use of metafounders or the algorithm for proven and young animals1[J].J Anim Sci,2020,98(3):skaa019.
[22]	HARDER I,STAMER E,JUNGE W,et al.Estimation of genetic parameters and breeding values for feed intake and energy balance using pedigree relationships or single-step genomic evaluation in Holstein Friesian cows[J].J Dairy Sci,2020,103(3):2498-2513.
[23]	BOHLOULI M,ALIJANI S,NADERI S,et al.Prediction accuracies and genetic parameters for test-day traits from genomic and pedigree-based random regression models with or without heat stress interactions[J].J Dairy Sci,2019,102(1):488-502.
[24]	MOGHADDAR N,VAN DER WERF J H J.Genomic estimation of additive and dominance effects and impact of accounting for dominance on accuracy of genomic evaluation in sheep populations[J].J Anim Breed Genet,2017,134(6):453-462.
[25]	ESFANDYARI H,BIJMA P,HENRYON M,et al.Genomic prediction of crossbred performance based on purebred Landrace and Yorkshire data using a dominance model[J].Genet Sel Evol,2016,48(1):40.
[26]	LI W, ZHENG M, ZHAO G, et al. Identification of QTL regions and candidate genes for growth and feed efficiency in broilers[J]. Genet Sel Evol, 2008. 91(11):4414-4423.
[27]	BEGLI H E,WOOD B J,ABDALLA E A,et al.Genetic parameters for clutch and broodiness traits in turkeys (Meleagris Gallopavo) and their relationship with body weight and egg production[J].Poult Sci,2019,98(12):6263-6269.
[28]	DE KETELAERE B,GOVAERTS T,COUCKE P,et al.Measuring the eggshell strength of 6 different genetic strains of laying hens:techniques and comparisons[J].Br Poult Sci,2002,43(2):238-244.
[29]	SAVEGNAGO R P,CAETANO S L,RAMOS S B,et al.Estimates of genetic parameters,and cluster and principal components analyses of breeding values related to egg production traits in a White Leghorn population[J].Poult Sci,2011,90(10):2174-2188.
[30]	康相涛,宋素芳,李明,等.蛋鸡种蛋蛋重对孵化率和雏鸡生长发育的影响[J].中国家禽,2002,24(15):10-11,13.KANG X T,SONG S F,LI M,et al.The effects of egg weights of breeding layers on hatchability and development of chicks[J].China Poultry,2002,24(15):10-11,13.(in Chinese)
[31]	WU X,LUND M S,SUN D,et al.Impact of relationships between test and training animals and among training animals on reliability of genomic prediction[J].J Anim Breed Genet,2015,132(5):366-375.
[32]	HABIER D,FERNANDO R L,DEKKERS J C M.The impact of genetic relationship information on genome-assisted breeding values[J].Genetics,2007,177(4):2389-2397.
[33]	GAO H,SU G,JANSS L,et al.Model comparison on genomic predictions using high-density markers for different groups of bulls in the Nordic Holstein population[J].J Dairy Sci,2013,96(7):4678-4687.
[34]	SU G,BRØNDUM R F,MA P,et al.Comparison of genomic predictions using medium-density (~54,000) and high-density (~777,000) single nucleotide polymorphism marker panels in Nordic Holstein and Red Dairy Cattle populations[J].J Dairy Sci,2012,95(8):4657-4665.
[35]	HABIER D,FERNANDO R L,KIZILKAYA K,et al.Extension of the Bayesian alphabet for genomic selection[J].BMC Bioinformatics,2011,12:186.
[36]	CLARK S A,VAN DER WERF J.Genomic best linear unbiased prediction (gBLUP) for the estimation of genomic breeding values[M]//GONDRO C,VAN DER WERF J,HAYES B.Genome-Wide Association Studies and Genomic Prediction.Totowa:Humana Press,2013:321-330.
[37]	LEGARRA A,AGUILAR I,MISZTAL I.A relationship matrix including full pedigree and genomic information[J].J Dairy Sci,2009,92(9):4656-4663.
[38]	DUDOIT S,VAN DER LAAN M J,POLLARD K S.Multiple testing.Part I.Single-step procedures for control of general type I error rates[J].Stat Appl Genet Mol Biol,2004,3(1):Article13.
[39]	SU G S,CHRISTENSEN O F,OSTERSEN T,et al.Estimating additive and non-additive genetic variances and predicting genetic merits using genome-wide dense single nucleotide polymorphism markers[J].PLoS One,2012,7(9):e45293.
[40]	KUMAR S, MOLLOY C, MUN~OZ P, et al. Genome-enabled estimates of additive and nonadditive genetic variances and prediction of apple phenotypes across environments[J]. G3 (Bethesda), 2015,5(12):2711-2718.
[41]	BAI H,SUN Y,LIU N,et al.Genome-wide detection of CNVs associated with beak deformity in chickens using high-density 600K SNP arrays[J].Anim Genet,2018,49(3):226-236.
[42]	TRIBOUT T,LARZUL C,PHOCAS F.Efficiency of genomic selection in a purebred pig male line[J].J Anim Sci,2012,90(12):4164-4176.
[43]	FORNI S,AGUILAR I,MISZTAL I.Different genomic relationship matrices for single-step analysis using phenotypic,pedigree and genomic information[J].Genet Sel Evol,2011,43(1):1.
[44]	LILLEHAMMER M,MEUWISSEN T H E,SONESSON A K.Genomic selection for two traits in a maternal pig breeding scheme[J].J Anim Sci,2013,91(7):3079-3087.
[45]	GUO X,CHRISTENSEN O F,OSTERSEN T,et al.Improving genetic evaluation of litter size and piglet mortality for both genotyped and nongenotyped individuals using a single-step method[J].J Anim Sci,2015,93(2):503-512.
[46]	YAN Y Y,WU G Q,LIU A Q,et al.Genomic prediction in a nuclear population of layers using single-step models[J].Poult Sci,2018,97(2):397-402.