Comparison of artificial intelligence algorithms and their ranking for the prediction of genetic merit in sheep.

As the amount of data on farms grows, it is important to evaluate the potential of artificial intelligence for making farming predictions. Considering all this, this study was undertaken to evaluate various machine learning (ML) algorithms using 52-year data for sheep. Data preparation was done before analysis. Breeding values were estimated using Best Linear Unbiased Prediction. 12 ML algorithms were evaluated for their ability to predict the breeding values. The variance inflation factor for all features selected through principal component analysis (PCA) was 1. The correlation coefficients between true and predicted values for artificial neural networks, Bayesian ridge regression, classification and regression trees, gradient boosting algorithm, K nearest neighbours, multivariate adaptive regression splines (MARS) algorithm, polynomial regression, principal component regression (PCR), random forests, support vector machines, XGBoost algorithm were 0.852, 0.742, 0.869, 0.915, 0.781, 0.746, 0.742, 0.746, 0.917, 0.777, 0.915 respectively for breeding value prediction. Random forests had the highest correlation coefficients. Among the prediction equations generated using OLS, the highest coefficient of determination was 0.569. A total of 12 machine learning models were developed from the prediction of breeding values in sheep in the present study. It may be said that machine learning techniques can perform predictions with reasonable accuracies and can thus be viable alternatives to conventional strategies for breeding value prediction.

Estimation of (Co)Variance components and genetic parameters of fibre traits in Rambouillet sheep using multi-trait analysis.

The present study aimed to estimate the genetic parameters of different fibre traits, viz., greasy fleece weight, staple length, and fibre diameter in Rambouillet sheep population using a multi-trait animal model. Data, spanning over 10 years (1998-2007) and pertaining to fibre traits at first clip, were collected for a total of 4186 Rambouillet sheep maintained at an organized farm. (Co)Variance structure and genetic parameters were estimated using a multi-trait animal model. The genetic analysis of data was performed based on restricted maximum likelihood (REML) procedure using WOMBAT software. The model incorporated sex of lamb (i = 1, 2), year of birth (j = 1-10), season of birth (k = 1-2), and litter size (l = 1-2) as fixed effects while direct additive genetic and maternal genetic effects were included as random effects. The direct additive genetic heritability estimates were 0.120±0.034, 0.136±0.037, and 0.356±0.070 for greasy fleece weight, staple length, and fibre diameter, respectively. The maternal genetic heritability of all fibre traits under study was very low. Additive genetic correlation was positive and low between greasy fleece weight and staple length; and between staple length and fibre diameter. In conclusion, fibre diameter was moderately heritable which implies that selection may lead to moderate improvement in this trait. The results from the present study will help in formulating optimal breeding plans for improvement of fibre traits in Rambouillet sheep.

Multivariate quantitative genetic analysis of body weight traits in Corriedale sheep.

In the present study, an attempt was made to elucidate the genetic parameters for body weight traits of lambs from Corriedale sheep population at different ages. Data were collected from 6874 lambs born over a span of 49 years from 1969 to 2017. The traits under study included body weight at birth (BW), weaning (WW), 6 months of age (6MW), 9 months of age (9MW) and yearling stage (YW). Data were statistically analyzed using restricted maximum likelihood (REML) algorithm in WOMBAT program. A multi-variate animal model was fitted to the data incorporating season and period of lambing, sex of lamb and litter size as fixed effects. Variance and covariance components were estimated using the animal model after incorporating direct additive genetic effect of animal as random factor. Genetic and phenotypic correlations with corresponding standard errors were also estimated. The heritability estimates for BW, WW, 6MW, 9MW and YW were 0.130 ± 0.023, 0.300 ± 0.029, 0.292 ± 0.030, 0.191 ± 0.025 and 0.169 ± 0.024, respectively. The genetic correlation between different traits under study was high, except between BW and 9MW for which the estimate was moderate. Phenotypic correlation ranged from low to high for different trait combinations. Among different traits under study, only two traits showed moderate heritability i.e. WW and 6MW while heritability of other traits was low. Both these traits showed high correlation with all subsequent traits. Selection programme for Corriedale sheep should be based on WW which is expressed early in life and shall lead to moderate genetic response to selection.