Estimation of Genetic Correlation: Interpretation of Experiments Using Selectively Bred and Inbred Animals

Abstract

There is increasing interest in determining the extent to which multiple characters related to drug sensitivity are influenced by common genes. The principal method for testing for the existence of such genetic correlations has been examination of pairs of mouse or rat lines selectively bred for sensitivity or resistance to a single behavioral effect of a drug. When a pair of selected lines is found to differ significantly on some trait other than the one on which they were selected, it is commonly concluded that significant genetic correlation between the traits exists, implying the action of a common set of genes on the two responses. In addition, results from comparisons of lines of animals selected for trait X and tested for trait Y may be compared with results from lines selected for trait Y and tested for trait X. As the number of correlated responses in selected lines increases, it becomes more important to adhere to sensible, consensual guidelines for interpreting such line differences. The principles underlying phenotypic and genotypic correlational analyses with selected lines are discussed. A scheme is presented to allow standardization across laboratories of inferences about the relative strength of genetic association from experiments with selected lines. Statistical and practical experimental issues are addressed. Estimates of genetic correlations may also be derived from the correlation of mean trait values across a panel of inbred strains. Existing data have sometimes found estimates of genetic correlations made with one approach to be inconsistent with those estimated in other ways. Possible reasons for this are discussed. Finally, the relationship between phenotypic correlations and genetic corrlations is discussed. Phenotypic and genetic correlations for a pair of traits may differ widely, and may even be opposite in sign. Both are characteristic of the population from which they are sampled. Phenotypic correlations estimated within selected lines may change over time, as the additive genetic variance in the selected trait is exhausted. A specific example of this phenomenon is given.