Figure 2

Relationship between GEBV accuracy after imputation and the correlation between true and imputed genotypes when using GBLUP for T = 200 and Ne = 100 in the imputation training set when using Beagle. Data are the means for each of the different methods when selecting the training set for D = 50, 100 and 200 SNPs/Morgan and MAF ≥ 0.00, 0.05 and 0.10; squares, triangles and circles denote 200, 100 and 50 SNPs/Morgan, respectively and symbols in black, grey and white denote MAF ≥ 0.10, 0.05 and 0.00, respectively; regression lines within MAF by D subclasses are also shown; identification of the different selection methods [see in Additional file1: Table S2]; standard errors for accuracy of GEBV vary between 0.003 and 0.004.