Estimation & Prediction

Ancestral regression to parents for the genetic evaluation of beef cattle.

---

Abstract

The "ancestral regression to parents" (PAR) was formally introduced in 2022, in a paper published in Genetics Selection Evolution. It is a recursive and Markov model of genetic evaluation like the classic "animal model" though it allows for the use of genomics to refine the pedigreed relationships from animals of three consecutive generations. In essence, for each animal evaluated that has been genotyped two parental path coefficients related to its sire and dam are estimated. In case, its parents and - at least - one grandparent per parental phase are not genotyped, its breeding value is predicted using a similar algorithm as in the animal model, but the parental path coefficients are higher than 0.50 because they carry the information on the differential recombination inputs of the grandparental genomes in the gamete that the bull or the cow passed to the calf. The estimating equations for the breeding value of an individual involve the BV of both parents and all four grandparents that are absorbed into the parental path coefficients. The inverse covariance matrix of breeding values under PAR are functions only of the parental path coefficients. These parameters condense the information of the difference between multilocus and single locus probabilities of identity by descent", a difference that is called identity disequilibrium (ID), following theory developed by C. C. Cockerham and B.S. Weir in late sixties and early seventies. Previous research from my group found that variance in ID for multilocus or continuous identity by descent relationships is between 50 to 70% smaller than the variance of genomic relationships of single locus identity by descent and more than three times smaller than identity by state genomic relationships among half-sibs. In theory, the higher the ID the larger the additive variance that may be available for prediction purposes and the greater the true accuracy achieved. I will present the results of running the PAR in the animal breeding unit at the University of Buenos Aires - INPA CONICET for the Latin-American Brangus evaluation since 2020, and for the Argentinian Braford since 2021. The multiple trait BLUP-PAR model for growth has more than three million equations in Brangus that are processed in one hour and a half, and more than two million equations in an hour for Braford in a regular laptop. The running time also includes the calculus of true accuracies, which are obtained with the efficient inversion algorithm of Takahashi. The solutions are quite consistent between consecutive evaluations, especially for heifers and young bulls.

Keywords: 2026

How to Cite:

Cantet, R., Kambo, S. & Rodríguez, D., (2026) “Ancestral regression to parents for the genetic evaluation of beef cattle.”, World Congress on Genetics Applied to Livestock Production Digital Archive 2026(1): 2287380. doi: https://doi.org/10.31274/wcgalp.24283

Rights: 1