Considerations for Including Individual Animal Genetic Information in Analyses of Experimental Data.

The objective was to explore the potential impacts of including individual animal genetic information in the analyses of experimental data. With the increasing availability of genetic information (e.g., breed, sire, EPD) on animals utilized in research experiments, considerations should be given to the benefits and limitations of incorporating these data. Incorporating individual animal genetic information could increase power. Ignoring individual animal genetic information could result in baseline imbalances. Consideration needs to be given to the type and availability of genetic information on research animals. Information could be as simple as breed or breed of sire, but could also include sire, maternal grandsire, EPD, etc. Next, consideration should be given to how this data would be incorporated in the statistical model. One advantage of using breed, sire, or maternal grandsire as a categorical fixed effect is that those effects would be modeled the same for all dependent variables. Another approach would be to use continuous variables such as EPD as a covariate. Theoretically, an EPD should explain more variation in ADG than just breed or sire would explain. One of the challenges with this approach is identifying the most appropriate EPD for each dependent variable. This results in a different model for each response variable, and some dependent variable may not have an associated EPD available. Another issue is that EPD of young cattle are low accuracy and have inherent error associated with them. The potential benefits of including an EPD for ADG as a covariate in the analysis of treatment effects on ADG are enticing. However, the measurement error associated with the difference between a low accuracy EPD for ADG and true genetic potential for ADG presents problems and limitations. When analyzing experiments with individual animal as the experimental unit, individual animal breed, sire, or EPD is easily incorporated in the model. However, when pen is the experimental unit, it is not possible to include the genetic information on individual animals when the observational unit is a pen mean. If genetic information on individual animals is to be included in the model when pen is the experimental unit, individual animal observational units are required for dependent variables. Careful consideration should be given to benefits and limitations of incorporating individual animal genetic information in analyses of experimental data.