Fig. 4: Multi-scale resolution enhancement evaluation of soScope on a spatial DNA dataset from a mouse liver tumor tissue using slide-DNA-seq. | Nature Communications

Fig. 4: Multi-scale resolution enhancement evaluation of soScope on a spatial DNA dataset from a mouse liver tumor tissue using slide-DNA-seq.

From: Tissue characterization at an enhanced resolution across spatial omics platforms with deep generative model

Fig. 4

a H&E image of the liver metastasis tissue and its clonal annotations based on spatial DNA principal components (PCs) (“Methods” section). b Illustration of multi-scale resolution enhancement experiment. The entire tissue is divided into 60 × 60 square regions. Neighboring regions at increasing sizes are merged to generate spatial DNA profiles at progressively lower resolutions. Enhancement approaches are required to enhance low-resolution profiles back to their original resolution. c Comparison of enhancement approaches to enhance top 2 DNA PCs from low-resolution profiles at different scales. d Evaluation of recovery accuracy in multi-scale resolution enhancement experiment using Pearson correlation (left) and MSE (right). e Evaluation of tissue structure (\(k\)=3) and substructure (\(k\)=4) identification using DNA PCs enhanced from low-resolution profiles. Subpopulation identification is performed using \(k\)-means clustering. Left: ground truth tissue structures identified from original data. ARI adjusted Rand index. Source data are provided as a Source Data file.

Back to article page