Fig. 1: Workflow of the DeepTalk. | Nature Communications

Fig. 1: Workflow of the DeepTalk.

From: Deciphering cell–cell communication at single-cell resolution for spatial transcriptomics with subgraph-based graph attention network

Fig. 1

a DeepTalk takes the single-cell RNA sequencing (scRNA-seq) data and spatial transcriptomics (ST) data as input. b DeepTalk-Integration integrates scRNA-seq and ST data using attentional graph neural networks. By employing self-attention mechanisms to capture cell relationships within scRNA-seq or ST data, and utilizing cross-attention mechanisms to explore connections between scRNA-seq and ST data, DeepTalk-Integration generate a weight matrix that represents the optimal cell type proportions for each cell or spot. c The integration results of DeepTalk-Integration. It’s mainly contains correction of low-quality data for spatially measured genes, cell-type localization and single-cell deconvolution. d DeepTalk-CCC predicts the cell–cell communications using subgraph-based attentional graph neural networks. e DeepTalk-CCC offers visualization outputs for spatially-resolved intercellular communication at the single-cell level. The Tissue component in (a) was created with BioRender.com, released under a Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International license.

Back to article page