A genome-wide association analysis for body weight at 35 days measured on 137,343 broiler chickens

Background
Body weight (BW) is an economically important trait in the broiler (meat-type
chickens) industry. Under the assumption of polygenicity, a “large” number of genes with “small” effects is expected to control BW. To detect such effects, a large sample size is required in genome-wide association studies (GWAS). Our objective was to conduct a GWAS for BW measured at 35 days of age with a large sample size.
Methods
The GWAS included 137,343 broilers spanning 15 pedigree generations and 392,295 imputed single nucleotide polymorphisms (SNPs). A False Discovery Rate of 1% was adopted to account for multiple testing when declaring significant SNPs. A Bayesian ridge regression model was implemented, using AlphaBayes, to estimate the
contribution to the total genetic variance of each region harbouring significant SNPs (1 Mb up/downstream) and the combined regions harbouring non-significant SNP.
Results
GWAS revealed 25 genome regions harbouring 96 significant SNPs on 13 Gallus
gallus autosomes (GGA 1–4, 8,10-15, 19 and 27), with the strongest associations on GGA4 at 65.67-66.31 Mb (Galgal4 assembly). The association of these regions points to several strong candidate genes including: (i) growth factors (GGA1, 4, 8, and 13–14); (ii) leptin receptor overlapping transcript (LEPROT)/Leptin receptor (LEPR) locus (GGA8), and the STAT3/STAT5B locus (GGA27), in connection with the JAK/STAT signalling pathway; (iii) T-box gene (TBX3/TBX5) on GGA15 and CHST11 (GGA1) (both related to heart/skeleton development); and (iv) PLAG1 (GGA2). Combined, the 25 genome regions harbouring the 96 significant SNPs explained ~30% of the total genetic variance. The region harbouring significant SNPs that explained the largest portion of the total genetic variance (4.37%) was on GGA4 (~65.67–66.31 Mb).
Conclusions
To the best of our knowledge, this is the largest GWAS that has been conducted for BW in chicken to date. Despite the identified regions, which showed a strong
association with BW, a high proportion of genetic variance attributed to regions
harbouring non-significant SNP supports the hypothesis that the genetic architecture of BW35 is polygenic and complex. Our results also suggest that a large sample size will be required for future GWAS of BW35.