Using recursion to compute the inverse of the genomic relationship matrix

Abstract Text:

A traditional algorithm to invert the numerator relationship matrix is based on the observation that the conditional expectation for an additive effect of one animal given the effects of all other animals depends on the effects of its sire and dam only, each with a coefficient of 0.5.  With genomic relationships, such an expectation depends on all other genotyped animals, and the coefficients do not have any set value. For each animal, the coefficients plus the conditional variance can be called a genomic recursion. If such recursions are known, the mixed model equations can be solved without explicitly creating the inverse of the genomic relationship matrix. Several algorithms were developed to create genomic recursions. In an algorithm with sequential updates, genomic recursions are created animal by animal. That algorithm can also be used to update a known inverse of a genomic relationship matrix for additional genotypes. In an algorithm with forward updates, a newly computed recursion is immediately applied to update recursions for remaining animals. The computing costs for both algorithms depend on the sparsity pattern of the genomic recursions. An algorithm for proven and young animals assumes that the genomic recursions for young animals contain coefficients only for proven animals. Such an algorithm generates exact genomic EBV in GBLUP and is an approximation in single-step GBLUP. That algorithm has a cubic cost for the number of proven animals and a linear cost for the number of young animals. All algorithms were evaluated with a simulated data set of 1500 genotypes and ssGBLUP. In the algorithm with sequential updates, setting very small elements in recursions to zero resulted in little sparsity. Setting larger elements to zero caused large errors in G^-1 due to accumulation of errors.  However, this algorithm worked very well for inv(A₂₂), especially when the pedigree depth was limited. When complete recursions were computed and small elements were set to 0, the accuracy of   and GEBV’s was almost unaffected but the sparsity level was moderate. The sparsity level increased to > 60% when G was blended with 20% of . In all computations involving the algorithm for proven and young animals, the correlations of GEBV with those using the regular algorithm were > 0.99. The genomic recursions can provide new insight into genomic evaluation and possibly reduce costs of genetic predictions with extremely large numbers of genotypes. 

Keywords: Genomic selection, single-step GBLUP, efficiency