|
|
||||||||
1 Department of Genetics and Animal Breeding, Agricultural Academy, AI. Mickiewicza 24/28, 30-059, Krakow, Poland
2 Centre for Genetic Improvement of Livestock, University of Guelph, Guelph, ON, Canada N1G 2W1
Three methods of updating genetic evaluations from an animal model were compared with the method of using all data and pedigree information on the basis of accuracy and computing requirements. The proposed updating schemes were chosen to reduce computing costs while solving the mixed model equations by an indirect approach. Method 1 utilized observations from only the latest period along with pedigree information from all periods. Method 2 also utilized observations from only the latest period, but pedigree information was limited to only animals with data in the latest period. Both methods incorporated information on solutions and diagonal elements of mixed model equations from an analysis of data in previous periods. Method 3 used only data and pedigree information from the latest four periods. Results were compared with those from the ALL method, an analysis of all data and pedigree information from all periods. Method 1 was nearly the same as the ALL method in accuracy and in computing requirements, and, therefore, adoption of method 1 would not be warranted under the assumed model. Methods 2 and 3 were less accurate but required much less computing effort than the ALL method. None of the three updating schemes was deemed acceptable as a replacement to an analysis of all available data and pedigree information.
Key Words: updating prediction animal models
Submitted on November 8, 1990
Accepted on January 22, 1991
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |