Skip to main content

Advertisement

Fig. 2 | Genome Biology

Fig. 2

From: A multi-task graph-clustering approach for chromosome conformation capture data sets identifies conserved modules of chromosomal interactions

Fig. 2

Hi-C clusters for a human embryonic stem cell (ESC) from spectral clustering (Spearman’s >0). a Heat map of Spearman’s correlation matrix of contact counts of human ESC, reordered according to spectral cluster assignments. b Bar plot showing the distribution of chromosomes in the clusters. The height of the bars corresponds to the size of the clusters. Colors and numbers correspond to different chromosomes. c Bar plot showing the distribution of clusters in the chromosomes. The height of the bars corresponds to the size of the chromosomes. Colors and numbers correspond to different clusters. d Different regulatory features grouped according to the spectral cluster assignments. We sorted each feature in each column separately to show better the enrichment of the features in the clusters. The features were standardized using z score. e − log10 of KS test P values. For each signal in each cluster, we compared the signal values inside and outside the cluster using the KS test to check if the values inside the cluster are significantly higher than the values outside the cluster. Note that for visualization, clusters were reordered based on their enrichment patterns to put clusters with similar patterns close to each other. f The same standardized feature matrix, clustered using k-means. g Enrichment ratio between spectral clusters and k-means clusters. The ratio between cluster c i of spectral clustering and c j of k-means was defined as (o/N)/(K/M) where M is the total number of bins, K is the number of bins in c j , N is the number of bins in c i , and o is the number of bins shared between c i and c j . 1D one-dimensional, ESC embryonic stem cell, KS Kolmogorov–Smirnov

Back to article page