Genomic selection(GS) as a promising molecular breeding strategy has been widely implemented and evaluated for plant breeding, because it has remarkable superiority in enhancing genetic gain, reducing breeding time an...Genomic selection(GS) as a promising molecular breeding strategy has been widely implemented and evaluated for plant breeding, because it has remarkable superiority in enhancing genetic gain, reducing breeding time and expenditure, and accelerating the breeding process. In this study the factors affecting prediction accuracy(rMG) in GS were evaluated systematically, using six agronomic traits(plant height, ear height, ear length, ear diameter,grain yield per plant and hundred-kernel weight) evaluated in one natural and two biparental populations. The factors examined included marker density, population size, heritability,statistical model, population relationships and the ratio of population size between the training and testing sets, the last being revealed by resampling individuals in different proportions from a population. Prediction accuracy continuously increased as marker density and population size increased and was positively correlated with heritability; rMGshowed a slight gain when the training set increased to three times as large as the testing set. Low predictive performance between unrelated populations could be attributed to different allele frequencies, and predictive ability and prediction accuracy could be improved by including more related lines in the training population. Among the seven statistical models examined, including ridge regression best linear unbiased prediction(RR-BLUP), genomic BLUP(GBLUP), Bayes A, Bayes B, Bayes C, Bayesian least absolute shrinkage and selection operator(Bayesian LASSO), and reproducing kernel Hilbert space(RKHS), the RKHS and additive-dominance model(Add + Dom model) showed credible ability for capturing non-additive effects, particularly for complex traits with low heritability. Empirical evidence generated in this study for GS-relevant factors will help plant breeders to develop GS-assisted breeding strategies for more efficient development of varieties.展开更多
There are two fundamental goals in statistical learning： identifying relevant predictors and ensuring high prediction accuracy. The first goal, by means of variable selection, is of particular importance when the tru...There are two fundamental goals in statistical learning： identifying relevant predictors and ensuring high prediction accuracy. The first goal, by means of variable selection, is of particular importance when the true underlying model has a sparse representation. Discovering relevant predictors can enhance the performance of the prediction for the fitted model. Usually an estimate is considered desirable if it is consistent in terms of both coefficient estimate and variable selection. Hence, before we try to estimate the regression coefficients β , it is preferable that we have a set of useful predictors m hand. The emphasis of our task in this paper is to propose a method, in the aim of identifying relevant predictors to ensure screening consistency in variable selection. The primary interest is on Orthogonal Matching Pursuit（OMP）.展开更多
利用微波辐射传输模型PWR(P. W. Rosenkranz)和反向传播神经网络方法,分别构建了正演下行辐射亮温和反演大气相对湿度廓线的模型,并研究了晴空条件下高光谱微波辐射计反演大气相对湿度廓线的通道选择问题。研究结果表明,200个通道的信...利用微波辐射传输模型PWR(P. W. Rosenkranz)和反向传播神经网络方法,分别构建了正演下行辐射亮温和反演大气相对湿度廓线的模型,并研究了晴空条件下高光谱微波辐射计反演大气相对湿度廓线的通道选择问题。研究结果表明,200个通道的信息含量大于微波辐射计7个通道的信息含量;增加探测通道数量可提升大气相对湿度廓线的反演精度,选取信息含量排在前面的120个通道进行仿真时,在0～2 km和6～10 km高度范围内大气相对湿度廓线的反演精度提升了4%～10%,在2～6 km高度范围内的相对湿度廓线的反演精度提升了约10%;当通道数继续增加时,反演精度的提升并不明显。展开更多
基金supported by the National Basic Research Program of China(2014 CB138206)National Key Research and Development Program of China(2016YFD0101803)+3 种基金the National Natural Science Foundation of China-CGIAR International Collaborative Program(31361140364)the Agricultural Science and Technology Innovation Program(ASTIP)of CAASFundamental Research Funds for Central Non-Profit of Institute of Crop Sciences,CAAS(1610092016124)supported by the Bill and Melinda Gates Foundation and the CGIAR Research Program MAIZE
文摘Genomic selection(GS) as a promising molecular breeding strategy has been widely implemented and evaluated for plant breeding, because it has remarkable superiority in enhancing genetic gain, reducing breeding time and expenditure, and accelerating the breeding process. In this study the factors affecting prediction accuracy(rMG) in GS were evaluated systematically, using six agronomic traits(plant height, ear height, ear length, ear diameter,grain yield per plant and hundred-kernel weight) evaluated in one natural and two biparental populations. The factors examined included marker density, population size, heritability,statistical model, population relationships and the ratio of population size between the training and testing sets, the last being revealed by resampling individuals in different proportions from a population. Prediction accuracy continuously increased as marker density and population size increased and was positively correlated with heritability; rMGshowed a slight gain when the training set increased to three times as large as the testing set. Low predictive performance between unrelated populations could be attributed to different allele frequencies, and predictive ability and prediction accuracy could be improved by including more related lines in the training population. Among the seven statistical models examined, including ridge regression best linear unbiased prediction(RR-BLUP), genomic BLUP(GBLUP), Bayes A, Bayes B, Bayes C, Bayesian least absolute shrinkage and selection operator(Bayesian LASSO), and reproducing kernel Hilbert space(RKHS), the RKHS and additive-dominance model(Add + Dom model) showed credible ability for capturing non-additive effects, particularly for complex traits with low heritability. Empirical evidence generated in this study for GS-relevant factors will help plant breeders to develop GS-assisted breeding strategies for more efficient development of varieties.
文摘There are two fundamental goals in statistical learning： identifying relevant predictors and ensuring high prediction accuracy. The first goal, by means of variable selection, is of particular importance when the true underlying model has a sparse representation. Discovering relevant predictors can enhance the performance of the prediction for the fitted model. Usually an estimate is considered desirable if it is consistent in terms of both coefficient estimate and variable selection. Hence, before we try to estimate the regression coefficients β , it is preferable that we have a set of useful predictors m hand. The emphasis of our task in this paper is to propose a method, in the aim of identifying relevant predictors to ensure screening consistency in variable selection. The primary interest is on Orthogonal Matching Pursuit（OMP）.
