MOESM4 of Contribution of rare and low-frequency whole-genome sequence variants to complex traits variation in dairy cattle

Additional file 4: Table S4. The Akaike information criterion (AIC) for different models compared to REML-PED. The log likelihoods for different models were used to calculate AIC following [35]. The AIC using “REML-PED” was scaled to zero for each trait and the AIC for the other models were expressed as the difference from AIC in “REML-PED”. GREML-MS” is the relative residual variance and DRP variance calculated using the GREML-MS method with partitioning of imputed sequence variants into MAF groups. “REML-GRM” is the relative residual variance and DRP variance calculated by fitting 50 k SNPs with the REML-GRM model implemented in GCTA. “REML-PED” is the relative residual variance and DRP variance calculated by fitting pedigree relationships with the REML-PED model implemented in DMU. “REML-PEDGRM” is the relative residual variance and DRP variance calculated by fitting both 50 k SNPs and pedigree relationships with the REML-PEDGRM model implemented in DMU. Results for the model that did not converge for the trait are not presented