Abstract
This study evaluated the accuracy and bias of estimated breeding values (EBV) using pedigree-based BLUP (PBLUP) and single-step genomic BLUP (ssGBLUP). Genomic evaluations for body weight (BW) and average daily gain (ADG) were compared using either all SNPs or a subset with minor allele frequency (MAF) 0.4-0.5. Data included 312 F2 broilers genotyped with a 60K Illumina BeadChip and 176 non-genotyped birds (total 488). The effect of reduced SNP density on prediction accuracy and bias for BW and ADG at 2-4 weeks of age was assessed using ssGBLUP. To examine the effect of reducing SNP density by changing minor allele frequency, SNPs with allele frequencies of 0.4-0.5 were isolated. The accuracy and bias of genomic predictions from the 0.4-0.5 MAF SNP subset were compared with those obtained using the standard 60K SNP array and traditional BLUP. Our study showed that using low-coverage genotype data would be a cost-effective approach for genomic prediction in crossbred chickens even with a small population size (less than 500).