Fig. 1

Prediction accuracy of genomic estimated breeding values from algorithm for proven and young single-step GBLUP (APY ssGBLUP) with different core definitions and core sizes for average daily gain in nursery (ADGn), average daily gain in finishing (ADGf), and backfat (BF). Core sizes were determined based on the number of largest eigenvalues explaining 50% (n = 160), 80% (n = 700), 90% (n = 1363), 95% (n = 2344), 99% (n = 7320) and > 99% (n = 10,000) of the genomic relationship matrix. The solid black line is the prediction accuracy estimated using regular ssGBLUP. MPA: most popular animals, both parents known and with own performance record; Rnd: a random subset of animals is sampled from all genotyped animals; Ped: animals were evenly selected from the pedigree sorted in genealogical order; Fam: from each full sib family, one progeny was allowed to be in the core group; Unrel: unrelated animals based on the genomic relationship were assigned to the core group; LPM: males with both parents known, without progeny, and without own performance records were used as core; LPF: females with both parents known, without progeny and own performance records were used as core; QR: animals selected based on a QR decomposition of the gene content matrix