Skip to main content

Table 2 Comparison of SBV predicted using longitudinal phenotypesa and EBV predicted using summarized phenotypesb, using pedigree and pedigree plus genomic information for the low (LRFI) and high (HRFI) RFI lines

From: Quality of breeding value predictions from longitudinal analyses, with application to residual feed intake in pigs

 

Pedigree

Pedigree + genomics

LRFI

HRFI

LRFI

HRFI

\(Bias\) \({\overline{SBV} }_{w}-\overline{{u }_{ySBVw}}\)

    

 \({SBV}_{1}\)

0.00

0.02

0.00

0.02

 \({SBV}_{2}\)

0.03

0.00

0.02

0.00

 \({SBV}_{3}\)

0.01

0.00

0.01

0.00

Dispersion \({b}_{{SBV}_{w},{u}_{ySBVw}}\)

    

 \({SBV}_{1}\)

1.05

1.03

1.04

1.03

 \({SBV}_{2}\)

1.03

1.02

1.03

1.03

 \({SBV}_{3}\)

1.00

1.01

0.97

1.00

Ratio of accuracies \({\rho }_{{SBV}_{w},{u}_{ySBVw}}\)

    

 \({SBV}_{1}\)

1.00

1.00

1.00

1.00

 \({SBV}_{2}\)

1.00

0.99

0.99

0.99

 \({SBV}_{3}\)

0.98

0.98

0.98

0.98

  1. aLongitudinal phenotypes = weekly measurements of FI analyzed with an RR model: \({FI}_{ij}={\mathbf{x}}_{ij}{\varvec{\upbeta}}+{{\beta }_{1}MBW}_{ij}+{\beta }_{2}{ADG}_{ij}+{\beta }_{3}{BFT}_{ij}+\sum_{q=0}^{2}{\mathrm{a}}_{iq}{\varphi }_{q}\left(j\right)+\sum_{q=0}^{2}{\mathrm{b}}_{iq}{\varphi }_{q}\left(j\right)+{e}_{ij}\), \({\mathrm{SBV}}_{\mathrm{l},\mathrm{i}}={\mathbf{L}}_{\mathrm{Gl}}{\mathbf{u}}_{\mathrm{i}}\), where \({\mathbf{L}}_{\mathrm{Gl}}\) is the \({l}^{th}\) eigenvector from the decomposition of the \(\mathbf{G}\) matrix
  2. bSummarized phenotypes:\({\mathbf{y}}_{{\mathbf{S}\mathbf{B}\mathbf{V}}_{\mathbf{l}}}={\mathbf{L}}_{\widehat{\mathbf{G}},\mathbf{l}}^{\mathbf{^{\prime}}}\left(\mathbf{F}\mathbf{I}-\mathbf{X}\widehat{{\varvec{\upbeta}}}-\widehat{{\beta }_{1}}\mathbf{M}\mathbf{B}\mathbf{W}-\widehat{{\beta }_{2}}\mathbf{A}\mathbf{D}\mathbf{G}-\widehat{{\beta }_{3}}\mathbf{B}\mathbf{F}\mathbf{T}\right)\)