Fig. 7
From: A benchmark study of deep learning-based multi-omics data fusion methods for cancer
C-index, silhouette score, Davies Bouldin score, and the association of the embeddings with survival and clinical annotations for the ten unsupervised methods on cancer multi-omics datasets. C-index (a), silhouette score (b) and Davies Bouldin score (c) were calculated based on the clustering on the cancer data. The number of clusters is set from two to six. The k-means clustering was run over 1000 times. d The embeddings which had strong association with survival (the Bonferroni-corrected p-values smaller than 0.05). The X-axis represents the number of survival-associated embeddings. The Y-axis represents cancers, and every cancer is assigned a color. e Selectivity score of the ten unsupervised methods for ten different cancer types. The score is displayed if it is higher than the average score (0.49), and the higher the selectivity score, the brighter the orange block