Modeling heterogeneous (co)variances from adjacent-SNP groups improves genomic prediction for milk protein composition traits

Table 3 Prediction reliability from univariate and bivariate BayesAS models

Trait^a	BayesAS-1SNP		BayesAS-100SNP		BayesAS-Genome
Trait^a	MT	ST	MT	ST	MT	ST
\({\upalpha}\) _S1-CN	0.10	0.09	0.13	0.09	0.10	0.09
\({\upalpha}\) _S1-CN-8P	0.04	0.02	0.06	0.03	0.03	0.03
\({\upalpha}\) _S2-CN	0.03	0.03	0.18	0.16	0.03	0.03
\({{\upkappa}}\)-CN	0.38	0.37	0.68	0.63	0.16	0.16
G-\({{\upkappa}}\)-CN	0.41	0.39	0.76	0.70	0.13	0.14
\({\upalpha}\)-LA	0.11	0.09	0.14	0.14	0.11	0.11
\({{\upbeta}}\)-LG	0.39	0.39	0.52	0.50	0.21	0.19
Protein %	0.14	0.14	0.18	0.17	0.12	0.11

^aProtein composition expressed as a fraction of the total milk protein percentage by weight wt (wt/wt), protein % expressed as percentage of the total milk yield; individual proteins comprise only the peaks identified as intact proteins and isoforms,i.e., \({\upalpha}\) _S1-CN (comprises \({\upalpha}\) _S1-CN 8P + 9P), \({\upalpha}\) _S2-CN (comprises \({\upalpha}\) _S2-CN 11P + 12P), \({{\upkappa}}\)-CN (comprises \({{\upkappa}}\)-CN G 1P + unglycosylated \({{\upkappa}}\)-CN 1P), where P = phosphorylated serine group. G-\({{\upkappa}}\)-CN = glycosylated-\({{\upkappa}}\)-CN; \({\upalpha}\) _S1-CN-8P = \({\upalpha}\) _S1-CN with 8 phosphorylated serine groups

ISSN: 1297-9686