Skip to main content
Fig. 1 | BMC Bioinformatics

Fig. 1

From: Snapshot: a package for clustering and visualizing epigenetic history during cell differentiation

Fig. 1

Overview of Snapshot. A Step1: cCRE indexing. A binarized index is created for each cCRE based on the presence/absence pattern of the cCRE across all cell types. B Step2: cCRE clustering and Step3: filtering. The cCREs with the same index were clustered into an Index-Set (IS). For example, the cCREs with 0_0_0_1_0 index were clustered into the IS in blue dash box. The cCREs in the less abundant ISs, highlighted by a black dash box, were filtered. C Step4: cCREs rescuing. The cCREs in the filtered ISs were re-classified as members of the abundant ISs based on their posterior probabilities of multivariate Gaussian distributions (using a Quadratic Discriminant Analysis (QDA) model) of the abundant ISs. The heatmaps in panels B and C were generated by deeptools [50]. D The mean signal matrix for all 68 abundant Index-Sets and an additional Index-Set, which included all remaining cCREs not assigned to an abundant IS. E The cCRE mean signal heatmap for the 19 Meta-Index-Sets (Meta-ISs) merged from 69 ISs. The number of Meta-ISs are automatically determined by AIC. F The bar plot for the number of cCRE within each Meta-ISs in log scale. G The frequency at which the signal pattern of a of Meta-IS was observed in 100 rounds of K-means clustering

Back to article page