Skip to main content

Advertisement

Table 5 Prediction mean squared errors for body mass index analysis by family layouts and genotype masking rates

From: Effect of genotype imputation on genome-enabled prediction of complex traits: an empirical study with mice data

90% genotype masking rate
Model* Across families Within families
1809 1809ia 201 1809 1809ia 201
BL 0.002 0.002 0.002 0.002 0.002 0.002
RKHS 0.002 0.002 0.002 0.002 0.002 0.002
BRANN 0.013 0.014 0.010 0.042 0.036 0.024
75% genotype masking rate
Model* Across families Within families
1809 1809ib 453 1809 1809ib 453
BL 0.002 0.002 0.002 0.002 0.002 0.002
RKHS 0.002 0.002 0.002 0.002 0.002 0.002
BRANN 0.021 0.023 0.015 0.044 0.045 0.041
50% genotype masking rate
Model* Across families Within families
1809 1809ic 905 1809 1809ic 905
BL 0.002 0.002 0.002 0.002 0.002 0.002
RKHS 0.002 0.002 0.002 0.002 0.002 0.002
BRANN 0.021 0.023 0.016 0.040 0.040 0.047
  1. aImputed from 201 SNPs.
  2. bImputed from 453 SNPs.
  3. cImputed from 905 SNPs.
  4. *BL: Bayesian LASSO; RKHS: Reproducing Kernel Hilbert Spaces (RKHS) and; BRANN: Bayesian Regularized Neural Networks.