Machine Learning Predicts the Oxidative Stress Subtypes Provide an Innovative Insight into Colorectal Cancer

Figure 2

NMF cluster samples into two oxidative stress subtypes. Before batch correction, three datasets are scattered (a). After batch correction, the component data is evenly distributed (b). The OS-related DEGs are selected by OS-related genes and DEGs (c). was best cut off by NMF analysis (d). PCA results show that two subtypes are grouped distinctly (e). Survival analysis demonstrates C2 patients with a poor prognosis, compared with C1 patients (f).

(a)

(b)

(c)

(d)

(e)

(f)