Table 2.
Whole genome methods | |||
---|---|---|---|
Method | Mean | s.d. | CI (95%) |
E-BLUP | 0.11 | 0.13 | (-0.13, 0.35) |
Bayes A | 0.27 | 0.12 | (0.04, 0.49) |
RKHS | 0.27 | 0.12 | (0.03, 0.50) |
Information gain using 2 classes (400 pre-selected SNPs) + RKHS | |||
Quantile | Mean | s.d. | CI (95%) |
0.15 | 0.33 | 0.12 | (0.09, 0.56) |
0.20 | 0.32 | 0.11 | (0.10, 0.53) |
0.25 | 0.36 | 0.11 | (0.13, 0.57) |
0.30 | 0.19 | 0.12 | (-0.05, 0.42) |
0.35 | 0.35 | 0.11 | (0.12,0.55) |
0.40 | 0.33 | 0.11 | (0.10, 0.53) |
Information gain using 3 classes (400 pre-selected SNPs) + RKHS | |||
Quantile | Mean | s.d. | CI (95%) |
0.15 | 0.32 | 0.11 | (0.10, 0.54) |
0.20 | 0.24 | 0.13 | (-0.01, 0.48) |
0.25 | 0.39 | 0.11 | (0.16, 0.59) |
0.30 | 0.19 | 0.12 | (-0.05, 0.42) |
0.35 | 0.20 | 0.12 | (-0.04, 0.43) |
0.40 | 0.16 | 0.12 | (-0.08, 0.40) |
E-BLUP: Bayesian linear model; Bayes A: Bayesian regression on SNP; RKHS: reproducing kernel Hilbert spaces regression