Reovirus antibody patterns in dogs: a trial for the application of principal component analysis to seroepidemiology.

In 100 dogs in Morioka and its vicinity serologically surveyed for reovirus infection, there were significant correlations between hemagglutination-inhibiting and neutralizing antibody titers of the same virus types. In order to establish a proper index for the evaluation of infection, principal component analysis was applied to the analysis of data, including standard immune sera. Of nine samples of matrix examined, correlation matrix of 6 variables was suggested to afford the most proper result. Essential information from the original data was accounted for by factor loadings of the first 3 components. The 6 attributes of serum antibody were classified into three groups. In a scatter diagram serum samples were classified also into three major groups corresponding to the three reovirus types, and the serotypic pattern of infection was clearly visualized.