Skip to main content
. Author manuscript; available in PMC: 2021 Jan 11.
Published in final edited form as: Front Phys. 2020 Jun 17;8:203. doi: 10.3389/fphy.2020.00203

Figure 2.

Figure 2.

Structure of the proposed GCNN model. The model includes two parts: graph convolution and a fully connected output layer for classification. Input is 1D gene expression levels of TCGA samples and the adjacency matrix of genes ( input graph). The graph is then pooled into a single GCNN layer to be fed into the hidden and output layers.