Skip to main content
. 2019 Jun 28;10:2892. doi: 10.1038/s41467-019-10769-x

Fig. 1.

Fig. 1

Human ulcerative colitis transcriptional signature does not cluster patients into molecular profiles. a Schematic representation of the strategy used for patient group identification, in which four publicly available data sets were combined. Gene ranking was done using the most variable genes in the human data set, which were used for clustering analysis. b Sample dissimilarity heatmaps for visual analysis of clustering tendency (VAT), comparing the human data set using the top 100 variable genes. c t-SNE plot using the top 100 variable genes in the human data set. Each point represents a patient sample. d Hartigan’s Dip test for clustering tendency using all genes in the data set, the top 100 variable genes, the top 100 highly dispersed genes, or the top 100 leading genes in the principal components. e Bootstrapping analysis of hierarchical clustering, comparing the human data set using the top 100 variable genes in the human data set. Numbers in orange indicate the approximately unbiased (AU) p-value, shown as a percentage. AU closer to zero indicates a cluster with low stability. Boxplot represents the median (center line, second quantile), first, and third quantiles (box) and whiskers extending 1.5 times the interquartile range (IQR)