Fig. 1: TAPE workflow and clarification of adaptive stage. | Nature Communications

Fig. 1: TAPE workflow and clarification of adaptive stage.

From: Deep autoencoder for interpretable tissue-adaptive deconvolution and cell-type-specific gene analysis

Fig. 1

a TAPE takes scRNA-seq data from human or mouse and RNA-seq data from the homologous tissue as input, then performs the deconvolution as well as the prediction of cell-type-specific GEPs via a training stage and an adaptive stage. b Generation of the cell-type-specific GEPs has two separate modes. The first is the “high-resolution” mode: TAPE takes the RNA-seq data from one sample at a time as input and outputs the adapted cell-type-specific signature matrix for each sample. The second is the “overall” mode: TAPE takes all the RNA-seq data at one time as input and outputs one signature matrix adapted to all samples. n is the number of samples, m is the number of genes, k is the number of cell types.

Back to article page