Fig. 6: Comparison of detecting layer structures in DLPFC sections and heterogenous tumor tissue regions in human breast cancer section. | Nature Communications

Fig. 6: Comparison of detecting layer structures in DLPFC sections and heterogenous tumor tissue regions in human breast cancer section.

From: GNTD: reconstructing spatial transcriptomes with graph-guided neural tensor decomposition informed by spatial and functional relations

Fig. 6

a The comparison of spot clustering with all genes in the raw data and the imputed data by tensor-based models, GNTD, FIST, CoSTCo and DTD as well as AE models, SEDR, STAGATE, and GraphST at the best rank on all 12 DLPFC sections. b The comparison of spot clustering with only highly variable genes in the raw data and the imputed data by the compared methods at the best rank on all 12 DLPFC sections. In the boxplot in both (a) and (b), the center line, box limits and whiskers denote the median, upper and lower quartiles and 1.5 × interquartile range, respectively. c Ground-truth of tissue regions and the detected spatial domains by clustering (Upper panel) and the UMAP embeddings of the spots by highly variable genes (Lower panel) of the raw data and the imputed data by the compared methods at the best rank on the DLPFC section 151673. d Ground-truth of tissue regions and the detected spatial domains by clustering (Upper panel) and the UMAP embeddings of the spots by highly variable genes (Lower panel) of the raw data and the imputed data by the compared methods at the best rank on the human breast cancer section. Source data for (a) and (b) are provided as a Source Data file.

Back to article page