EFFECT AND TREATMENT OF MISCLASSIFIED SAMPLES IN LINKAGE ANALYSIS

When field data are used for genome analyses in farm animals, misclass ification can occur. This can be caused either by errors in the pedigr ee files, or by non-intentional exchanges of milk or blood samples whe n collecting and processing material for genotyping. A statistical mod el is suggested, which takes the probability of misclassification Delt a into account when estimating the recombination rate theta between tw o markers. These two parameters can be estimated jointly using a maxim um-likelihood approach. This approach is used to estimate the recombin ation rate between the two polymorphic milk-protein systems kappa-case in and beta-casein, from genotype data of Austrian Simmental (SI) and Austrian Braunvieh (BV). If the possibility of misclassification is di sregarded, the recombination rate between the two loci is estimated to be 0.18 and 0.06, respectively. However this contradicts the well-kno wn fact that these two loci are in very close linkage. With the full m odel, the estimated recombination rate was 0.054 in SI and 0.0 in BV, with both estimates not significantly different from zero. The corresp onding estimated misclassification rates were 0.12 and 0.14, respectiv ely, and were highly significant. If the possibility of misclassificat ion in genetic studies is ignored, the recombination rare will be syst ematically overestimated and misleading conclusions may result.