Skip to main content
Fig. 1 | BMC Genomics

Fig. 1

From: Modular combinatorial binding among human trans-acting factors reveals direct and indirect factor binding

Fig. 1

RMD discovers compact and comprehensive regulatory modules. a RMD discovers regulatory modules from regulatory regions across the genome. A region may use one or more modules. A factor may participate in one or more modules. RMD uses a topic model to summarize the binding data into modules and to assign binding sites to the modules. b The topic model re-capitulates the original binding data more accurately than k-means clustering and NMF. Each point in the scatter plot represents the Pearson correlation coefficient between a pair of TFs calculated using the original binding data (x-axis) or calculated using the reduced data matrix (k = 49) by topic model, k-means clustering, or NMF (y-axis). c A heatmap shows the Pearson correlations between topic model modules and k-means clusters (k = 49). The modules are ordered as in Fig. 2. Bottom bar chart shows the maximum correlation values for each k-means cluster. All k-means clusters are matched by at least one module, with a maximum correlation value larger than 0.5. Right bar chart shows the maximum correlation values for each module. Ten modules cannot be matched by any k-means clusters

Back to article page