Figure 1.

DCNet architecture and design process. (a) Different cell types in the tissue, such as stromal cells, red blood cells, macrophages, T cells, B cells, neuronal cells, etc. These cells can be further divided into different cell types, which has a corresponding marker gene. (b) The basic structure of the DCNet model. The input is the expression levels of marker genes in bulk RNA-seq, and the output is the expression levels of all genes. The middle layer artificial neural network. (c) The gray neurons are activated neurons, and the white ones are inactive neurons. The first level relationship of the DCNet model is the corresponding relationship between cells and marker genes. The middle layer of the DCNet model represents the relative content of cells.