Fig. 2: stClinic is able to align multiple SRT datasets across diverse samples. | Nature Communications

Fig. 2: stClinic is able to align multiple SRT datasets across diverse samples.

From: stClinic dissects clinically relevant niches by integrating spatial multi-slice multi-omics data in dynamic graphs

Fig. 2

a Manual annotation of the 12 slices across three samples on the human DLPFC dataset, including six (or four) layers and WM. b Bar plot illustrating clustering accuracy in terms of ARI on three samples by SEDR, GraphST, STitch3D, PRECAST, STAligner, stClinic_fix, and stClinic. c Spatial domains detected by SEDR, GraphST, STitch3D, PRECAST, STAligner, and stClinic on the Sample 1. d UMAP visualization of the latent features by SEDR, PRECAST, STAligner, stClinic_fix, and stClinic across all 12 slices. In the top and bottom panels, the colors represent the slices and clusters, respectively. e Comparison of the seven methods (SEDR, GraphST, STitch3D, PRECAST, STAligner, stClinic_fix, and stClinic) regarding accuracy of clustering (ARI and NMI) and batch-effect correction (cLISI and iLISI), on N = 12 slices. It’s noted that lower cLISI values indicate better correction for cell-type mixing, while higher iLISI values indicate better correction for batch mixing. For each boxplot, the center line, box limits, and whiskers separately indicate the median, upper and lower quartiles, and 1.5 × interquartile range. f Spatial domains identified on slice 151672 by five methods (SEDR, PRECAST, STAligner, stClinic_fix, and stClinic) under the condition of integrating 12 slices, respectively. Layer 3 is outlined in black. Source data are provided as a Source Data file.

Back to article page