Showing 2 results for Random Regression
H. Farhangfar, H. Naeemipour , R. Lotfi,
Volume 12, Issue 43 (4-2008)
Abstract
This study was undertaken to estimate genetic trend and parameters of Holstein cattle in Khorasan province for milk yield using a spline random regression test day animal model. A total of 32854 monthly test day milk records (twice and thrice a day milking) obtained from 3842 Holstein heifers (progeny of 466 sires) distributed in 125 herds and calved from 2001 to 2005 was used to predict breeding value of individual animals. In the model, fixed effects of herd including year-month of recording, milking times, age at calving (linear and quadratic covariables), Holstein gene percentage (linear covariable) as well as random effects of additive genetic and permanent environment were studied. To take account of the shape of the lactation curve at genetic and environmental levels, cubic spline polynomials were also included in the test day model. Bayesian method by applying Gibbs sampling technique (100000 chains applying RRGIBBS software) was utilized to obtain posterior means of predicted breeding value of animals for milk yield at individual month of lactation. The results showed that mean of breeding value for 305-day milk yield was 52.90 kg (p<0.05). Spearman rank correlations between predicted breeding values at different months of lactation decreased as the interval between them increased. The highest and lowest rank correlations were found between months 8 and 9 (0.998) and between months 1 and 10 (0.312), respectively. Predicted breeding value of 305-day milk had the lowest and highest rank correlations with predicted breeding value at months 1 (0.553) and 6 (0.990), respectively. Regression analysis of average predicted breeding value of progenies in their birth year showed that the amount of genetic trend for 305 day milk yield was 17.75 kg per year, statistically no different from zero (p value=0.165).
F Ghafori, M Eskandari, H Mohamadi,
Volume 13, Issue 47 (4-2009)
Abstract
Variance components and genetic parameters of body weight of Mehraban sheep were estimated by univariate and random regression models. This was done by using body weight records of 2746 Mehraban lambs related to flocks under supervision of the Agriculture Organization of the Hamadan province, collected between 1990 and 2005. In both methods, variance components estimates were obtained by restricted maximum likelihood (REML) using DFUNI and DXMRR programs, respectively, via DFREML 3.1 software package. Results showed that variance components obtained from RR models (except for residual variance) in some ages were higher than those obtained from univariate models. Direct heritability (h2) estimates from univariate and RR models were approximately equal to weaning age but, overall, RR estimates were higher than those obtained from univariate analyses. Maternal heritability estimates (m2) from RR models were higher than univariate models’ estimates, and showed a different pattern of variation with age. Correlations between predicted breeding values from univariate and RR models for birth weight and weaning weight were 0.72 and 0.70, respectively. Results showed that estimates of variance components and genetic parameters by RR models were affected by data structure and in case of the need for genetic parameters, especially those related to body weight late in lambs’ life, estimates of univariate analyses should be preferred.
