Skip to main content
Fig. 1 | Genome Biology

Fig. 1

From: ChromNet: Learning the human chromatin network from all ENCODE ChIP-seq data

Fig. 1

a Top: Interaction network among four simulated regulatory factors. Middle: Binding activity from simulated ChIP-seq data sets, where each peak represents a putative binding position of a protein. Bottom: Networks inferred from ChIP-seq data sets based on co-occurrence (left) or conditional dependence (right). b Comparison of separate cell-type networks (top) with a single joint network (bottom). In a joint model, factors in each cell type have opportunities to be connected with new regulatory factors in other cell types, as highlighted by the blue shaded region (bottom). c Redundant information obscures conditional-dependence connections. Left: Without redundancy, standard methods robustly infer a conditional-dependence network. Middle: Highly correlated variables (such as A and A ′) are strongly connected with each other and lose their connections with other variables. Right: A group graphical model (GroupGM) represents the conditional dependence between groups of correlated variables, which restores the connection between A and B

Back to article page