A case study on computer system for calculation of estimated breeding value for swine seedstock improvement

To improve performance level of a swine herd, genetically superior pigs must be selected. It is necessary to utilize not only an animal's own data but also relatives' data to increase the accuracy of prediction of the breeding value. In this paper the development of a computer program for calculating EBV's on economically important traits in swine and of selection indices based on the Japanese economic values is reported with reference to the Programming Outline for Calculating Estimated Breeding Values in Swine, Purdue University. In the program, EBV's are calculated on five traits: number of pigs born alive (NBA), litter weight at 21 days (LW 21), body weight at 56 day-old (WT 56), number of days required to reach to 105kg of body weight (DAY) and average back-fat thickness at the weight (BF). These estimates are based on an individual's and relatives' (progeny, parents, full-sibs and half-sibs) data and selection indices are calculated for boars, sows, off-test pigs and young boars. Three selection indices for 1) Female Line, 2) Rotational Line and 3) Terminal Sire Line are calculated on pigs off-test. For the calculation of EBV's, a large number of relatives' information in the data set are utilized. When using a small computer, it is necessary to calculate EBV's quickly and accurately. With the master data files of about 550 sows and 110 boars, it takes about 50 minutes to calculate EBV's of NBA and LW 21 with the contemporary group of about 60 sows. For EBV's of DAY and BF with approximately 600 off-test pigs in a group, only about 15 minutes are needed for the calculation. Also calculation of EBV's of WT 56 requires approximately one minute with about 40 pigs in a contemporary group.