Fig. 3 | Scientific Reports

Fig. 3

From: A machine learning model and molecular clusters of epigenetic chromatin regulators in tuberculosis based on bioinformatics and clinical samples

Fig. 3

Identification of molecular clusters in TB based on DE-CR expression profiles.(A) Consensus matrix heatmap indicating the stability of sample clustering when k = 2, with clearer block-like structures suggesting distinct clusters. (B) Cumulative distribution function (CDF) curves for consensus clustering. The minimal change in the CDF curve area between k = 2 and k = 3−9 (panel C) supports k = 2 as the optimal choice. (C) Relative change in area under the CDF curve for different k values. A negligible increase after k = 2 indicates that further clustering does not provide significant improvement. (D) Cluster-consistency plot indicating that the highest consensus scores for each subtype are achieved at k = 2. (E) Principal component analysis (PCA) plot visually confirming the clear separation between the two identified molecular clusters, Cluster 1 (n = 41) and Cluster 2 (n = 51).

Back to article page