Figure 1.
(A) The structure of the HiC-SGL model. It consists of an encoder and a decoder. The encoder extracts the local feature of each edge and the global feature of the cell graph. The decoder estimates the likelihood of each edge being present in the cell map using the encoded feature. (B) Subgraph Encoder, implemented by variant of Transformer on graph neural networks. : Node Embeddings of layer
; E: adjacency matrix of subgraph.