Skip to main content
. 2024 Mar 29;5(3):100287. doi: 10.1016/j.xhgg.2024.100287

Figure 6.

Figure 6

Assessment of the strength of the identified MKHK-ID4 episignature, cross-validation, and MVP scores

(A) Hierarchical clustering, with red representing ID4 case samples and blue depicting matched control samples in the heatmap panel.

(B) MDS, where red and blue circles represent case and control samples, respectively. Clear separation of the ID4 case samples and control samples is observed in plots A and B.

(C) Rounds of leave-one-out cross-validation were performed, using all but one ID4 case sample for probe selection at each trial. It was observed that, at all iterations, the black circle (ID4 sample not used for probe selection) clustered with the red ones (ID4 samples used for probe selection) and far from the blue circles (control individuals), demonstrating the sensitivity of the episignature.

(D) MVP scores created by the SVM constructed using the ID4 selected probes. All the ID4 case samples have received MVP scores near 1 and all the control samples and case samples from other disorders have received scores near 0, indicating full specificity of the model. All the other MKHK samples, other than individuals C.ZT.1 and C.ZT.2 (demonstrated by green and orange circle, respectively), have received low MVP scores. Blue circles represent training samples, while gray circles depict testing samples.