Table 4.

Performance of combined feature set 1.2.9.

Linear Regression	cg04692870-Probe 1	cg09322432-Probe 3	cg10840135-Probe 4	cg15597984-Probe 5	cg20046859-Probe 7	Mean (SD)
RMSE train set	0.065	0.021	0.025	0.032	0.017	0.0318 (0.0193)
R2 train set	0.167	0.294	0.217	0.230	0.258	0.233 (0.0473)
RMSE test set	0.069	0.025	0.029	0.028	0.022	0.0346 (0.0196)
R2 test set	−0.217	−0.403	−0.224	0.0747	−0.480	−0.250 (0.214)

GWAS of 75% samples after B-H correction, each of the 15 significant SNPs across all probe GWASes, mother’s income, household income, accommodation, mother’s highest education level, child’s ethnicity and child’s sex as independent predictors in the prediction of CpG loci methylation beta values for cg04692870-Probe 1, cg09322432-Probe 3, cg10840135-Probe 4, cg15597984-Probe 5, and cg20046859-Probe 7, using linear regression. Performance for each CpG site and average performance across five CpG sites used in prediction were reported.