Unsupervised clustering of metabolic enzyme activity profiles across 33 TCGA cancer types. NMF decomposes the iMetAct-inferred enzyme activity matrix to discover metabolic subtypes, revealing distinct pathway signatures, survival outcomes, and immune microenvironments for each cluster.
Select cancer type, database, and method to determine optimal cluster count from the NMF diagnostic plot.
Choose k based on the diagnostic, then generate metabolic pathway, survival, and immune profiling visualizations.
Select cancer type, pathway database, and NMF algorithm to load the diagnostic plot. Evaluate cophenetic correlation, silhouette scores, and dispersion to identify the best number of metabolic subtypes.
After choosing k, generate pathway enrichment heatmaps, Kaplan-Meier survival curves, tumor grade distribution, and CIBERSORT immune profiling in a single click. Download each plot individually.
Load the NMF diagnostic plot to assess clustering quality across k = 2–5. Choose the k where cophenetic correlation remains high and silhouette scores peak.
33 TCGA cancer types with pre-computed NMF results
KEGG: 186 pathways · Reactome: 1,442 pathways
brunet: standard divergence · lee: multiplicative updates
Configure parameters and run the analysis to see results here.