Fig. 4
From: Probabilistic modeling methods for cell-free DNA methylation based cancer classification
An example of the six first principal components from PCA and two ISPCA approaches for one of the data splits. The training and test samples from the discovery cohort were plotted with different colors to demonstrate how well the test samples will blend in with the training samples. A Components from PCA, using AML class DMRs as input. B Components from binary ISPCA, where class labels were set to 1 for the samples in AML class and to 0 for the samples from other classes. C Components from multiclass ISPCA