Figure 7. Consensus clustering analysis of CM patients based on the prognostic ARLs. (A) The consensus clustering heatmap shows the optimal classification when K = 2-9. (B) Principal component analysis clearly demonstrates a obvious distinction between the CM patients in Cluster A and Cluster B based on the prognostic ARLs. (C–G) The stromal, immune, ESTIMATE score and tumor purity of patients in Cluster A and Cluster B. (H) The fraction of 22-type immune cells of patients in Cluster A and Cluster B via CIBERSORT algorithm. (I) The fraction of 23-type immune cells of CM patients in Cluster A and Cluster B via the ssGSEA algorithm.