Moreover, all of the signifi cant SNP effects for DPR in this study were between 5 and 25 times greater than the largest marker effect from the BovineSNP50 chip. This result is probably due to the differences in done SNP selection between the two methods. The majority of SNPs on the BovineSNP50 chip are between genes and over 14,000 genes are not represented by a SNP on the Bovine SNP50 chip. In the current study, almost all of the SNPs examined were located within the coding region of the gene and the remainder were close physically to the coding region. Moreover, SNPs were chosen to maximize the probability that there would be a change in the characteristic of the protein encoded for the gene. Thus, it is likely that many of the SNPs that have large ef fects on DPR do so because they are causative SNPs resulting in changes in protein function.

The remainder may represent linkages to causative SNPs. The SNPs iden tified in this study may be closer to the causative SNPs than the SNPs on the BovineSNP50 chip. Allele substitu tion effects were estimated individually with a linear mixed model, rather than simultaneously as described in Cole et al. which also could explain some of the differences. Polymorphisms in the current study were chosen for having the greatest probability of changing protein func tion. In order to maximize the possibility of finding causative SNPs, we prioritized the selection of SNPs within a gene to favor those causing the greatest change in protein function. This decision may have been one reason why there was a high rate of SNPs with MAF 5% because the SNP would be subjected to puri fying selection.

Only 20% of the Entinostat nonsense, 25% of the missense and 9% of the frameshift mutations had MAF 5% whereas this frequency was 80% of the 5 SNPs that were in a non coding region or did not result in an amino acid substitution. Many of the SNPs were not in Hardy Weinberg equilibrium and this, too, may reflect the effect of the SNPs on protein function. Of the 9 SNPs most out of equilibrium, only 3 had less than expected fre quencies of minor allele homozygotes. The interpret ation is that few of the mutations in which MAF was 5% were lethal. Interestingly, for six genes, the heterozy gote was more or less frequent than expected. Some of the decrease in heterozygosity could be due to inbreeding, which is high in Holstein cattle.

Other changes in het erozygosity could be due to either an advantage or disad vantage of the heterozygote. Heterozygote advantage could be due to the ability of receptors to recognize more forms of the peptides they bind, heterozygotes having the optimal level of JAK1/2 inhibito gene expression, or in theory, the optimal allele being different for dif ferent cell types. A reason for heterozygote disadvantage is not clear. The antagonistic genetic relationship between fertility traits and milk production was verified here.