Technical note: updating the inverse of the genomic relationship matrix

A computing strategy to update the inverse of the genomic relationship matrix when new genotypes become available is described. It is shown that re-using results of previous computations can result in substantial reductions in computing time required. For instance, when the number of individuals increased by about 1% for matrices larger than 15,000, the time required for updating was less than 7% of that used for direct inversion from scratch.