Fig. 1: Imputation accuracy using the haplotype reference panel and its application to GWAS and breeding value estimation.
a Imputation accuracy using ten common commercial pigs (Duroc, N = 4; Landrace, N = 3; Large white, N = 3) as targets and the remaining 1854 haplotypes as reference panel. The x-axis shows the minor allele frequency of imputed variants. The y axis shows imputation accuracy measured by average R2 (squared Pearson correlation). The colored triangles represent the R2 under different minor allele frequencies. The 50k, 60k, 80k, 100k, and 300k denote the number of randomly selected polymorphism sites being used for imputation. b The manhattan plots of GWASs for Intramuscular fat (IMF) content before and after genotype imputation. The x-axis shows chromosomes. The y-axis shows significant levels of association. Each dot denotes a variant. Wald test is employed to estimate the significance of each variant by gemma v0.98 software. Reported log transformed P Values are two-sided and nominal (i.e. not corrected for multiple testing). The genome-wide significance threshold was set to 5 × 10-8. ENS40980, ENSSSCG00000040980; ENS50360, ENSSSCG00000050360. c The accuracy of estimated breeding value (EBV) before and after genotype imputation. Significant variants at the P < 0.01 level from GWAS were used for estimating the breeding value of IMF by BLUP. The x-axis shows the estimated breeding value. The y-axis shows the phenotypic value of IMF. Each dot denotes an individual. Pearson’s correlation is employed to evaluate the accuracy of EBV, indicated by an R2 (squared Pearson correlation) between phenotypic value and EBV. The P values of correlation coefficient before and after imputation are 5.2 × 10-222 and 2.7 × 10-358, respectively.
