Accuracy of whole genome prediction with single-step GBLUP in a Chinese yellow-feathered chicken population

Abstract

Single-step genomic best linear unbiased prediction (ssGBLUP) for genetic evaluation integrates phenotypes, genetic markers, and pedigree records into whole genome prediction simultaneously. ssGBLUP was expected to gain higher predictive ability than GBLUP since phenotypes of the historical non-genotyped individuals were included in the model. The objective of this study was to investigate the performance of ssGBLUP in genetic evaluation of a Chinese yellow-feathered chicken population. Predictive ability of BLUP, GBLUP, and ssGBLUP models were evaluated in both a random individual and a family cross-validation. Results showed that ssGBLUP outperformed the other models in both individual and family cross-validation scenarios in respect of average predictive ability. Moreover, a remarkable improvement of predictive ability from BLUP to ssGBLUP was observed (22.01 ± 17.70% of improvement on average), which reflect the benefit of implementing genetic markers, historical records, and pedigree into the genetic evaluation procedure. Our findings suggested that the single-step genetic evaluation procedure could be successfully implemented in the genomic selection of Chinese yellow-feathered chickens. The ssGBLUP model gave best predictions for most traits in the studied population and performed better in the individual cross-validation scenario than in the family cross-validation.