具有亲缘关系矩阵的混合模型的方差组分迭代估计——应用于计算BLUP的方差比值和遗传参数

论述的是来自非均街资料的混合模型中具有亲缘关系矩阵时利用迭代法估计方差组分问题。这篇文章表明计算程序是可行的,只要能够按照混合模型中固定效应的结构矩阵和Henderson方法3的固定效应的假设条件正确地计算二次型约化平方和,就可获得较为精确的方差组分估计值;而且表明方差初始比值k偏高或偏低,不影响迭代求解的最后结果,这是因为在迭代过程中可以通过结构矩阵x'x和x'x的控制而自行调整。这些方差组分不仅可应用于选种种畜用的BLUP计算,还可用来估计遗传参数。

Iterative Estimation of Variance Components for A Mixed Model with A Numerator Relationship Matrix: Apply to Compute K of Variance Ratio in BLUP and Genetic Parameters

An iterative procedure is presented for the estimation of variance components in the mixed model with a numerator relationship matrix from non-balanced data. This method has been shown to be feasible if only designing matrix for fixed effects of the mixed model and Henderson method 3 hypothetic condition of the fixed effect to compute correctly the reduction in sum of squares, we would obtain more precise estimation of variance components, furtermore, the sizes of the inital value K could not affect to last result of the iterative soluation. This is because in the Iterative process may adjust automally by designing matrix- Also, the variance components could apply not only to compute variance ratio in BLUP, but also to estimate genetic parame-tes.