2022 Volume 50 Issue 2 Pages 31-38
We performed Bayesian clustering analysis using STRUCTURE software with genotype data on 33,063 commercial single nucleotide polymorphism (SNP) markers in 4,348 Japanese Black fattened steers slaughtered at carcass markets in Tokyo, Osaka, Hyogo, Tottori, and Hiroshima prefectures. When the number of the assumed clusters in STRUCTURE was 2, the steers from Hyogo prefecture were clearly separated from the others. This indicates the usefulness of the STRUCTURE analysis with commercial SNP markers for the clarifications of the difference of the genetic constitutions of each prefecture. Next, genomic predictions for carcass traits were conducted using a statistical model including the proportions of the clusters as partial linear regressions. Genomic breeding values predicted by the model without the STRUCTURE covariates were likely to be divided into the part of explaining the STRUCTURE analysis and the remaining part. This result shows the possibility that the accuracy of genomic prediction depends on the degree of information of the genomic population structure.