Fuzzy classification of phantom parent groups in an animal model

Area of Science:

• Animal Genetics
• Quantitative Genetics
• Bioinformatics

Background:

• Genetic evaluation models commonly use genetic groups to address varying genetic levels in animals with unknown parentage.
• Traditional phantom parent groups often incorporate a time component, but combining time periods can lead to inaccuracies as all phantom parents are treated as contemporaries.

Purpose of the Study:

• To introduce and evaluate a fuzzy logic approach for classifying genetic groups in animal breeding.
• To improve the accuracy and reduce variability in the prediction of breeding values for animals with unknown parentage.

Main Methods:

• A fuzzy logic approach was developed where phantom parents can belong to multiple genetic groups with assigned proportions.
• Coefficients for the inverse of the relationship matrix were determined for fuzzy-classified genetic groups.
• Simulated data from ten generations of mass selection were used, with random deletion of observations and pedigree records. Phantom parent groups were defined by gender and generation, with simulated uncertainty in birth generation for some animals.

Main Results:

• Fuzzy-classified genetic groups resulted in slightly lower empirical prediction error variance (PEV).
• The ranking of animals with unknown parents was more accurate and less variable compared to distinct genetic groups.
• Assigning phantom parents to multiple groups based on gender and generation, with proportions reflecting true birth generation, further reduced empirical PEV.

Conclusions:

• Fuzzy classification offers a more parsimonious and structured method for describing the genetic level of unknown parents.
• This approach enhances the precision of predicted breeding values in genetic evaluations.