Skip to main content
. 2019 Mar 18;14(4):405–420. doi: 10.1080/15592294.2019.1588685

Table 2.

Mean R2 using the twofold and threefold cross-validation methods.

  2-fold cross-validation
3-fold cross-validation
Tissue pair All CpGs Removed CpGs with min methylation beta values >0.9 or max beta values <0.1 Removed CpGs with min methylation beta values >0.8 or max beta values <0.2 All CpGs Removed CpGs with min methylation beta values >0.9 or max beta values <0.1 Removed CpGs with min methylation beta values >0.8 or max beta values <0.2
measured Placenta-measured Cord Blood 0.6567879 0.6333165 0.5079859 0.6567879 0.6333165 0.5079859
measured Placenta- SVM predicted Placenta 0.9672072 0.9644117 0.9472045 0.9676109 0.9648506 0.9478592

For the ‘All CpGs’ column, we used all CpG sites in each sample to calculate the CpG-wise R2, and then obtained the mean R2. For the ‘Removed CpGs with min methylation beta values>0.9 or max beta values<0.1’ column, we excluded the extreme CpGs that fell within this range and used the remaining data to calculate the mean CpG-wise R2, and this column is similar to the ‘Removed CpGs with min methylation beta values>0.8 or max beta values<0.2’ column. For the ‘measured Placenta-measured Cord Blood’ row, the R2 is the squared correlation coefficient of measured methylation in both placenta and cord blood. For the ‘measured Placenta-SVM predicted Placenta’ row, the R2 is the squared correlation coefficient of measured methylation in the placenta and predicted methylation in the placenta by single-CpG-based SVM.