Figure 2.

Application of LSGI on single ST dataset.

A. Visualization of LSGI output on the spatial map (dataset: UKF243_T_ST from the RaviGBM study). Each rhombus represent a data spot (10x Visium technology) while the overlaying dark grey circles represent data spots characterized as tumor region. Each arrow indicate the presence of a gradient and the colors represent different NMF program of this gradient. Arrows directions indicate the direction of gradients.

B. Spatial proximity of different gradients. The colors represent the log-transformed distance from the NMF program in a row to the program in a column. Here the distance is the real physical distance. Notice that this matrix is not symmetric (Methods).

C-D. Visualization of the proximal NMF program pairs (C: NMF_2/4; D: NMF_3/5). Each arrow indicate the presence of a gradient and the colors represent different NMF program of this gradient. Arrows directions indicate the direction of gradients. The overlaying dark grey circles represent data spots characterized as tumor region.

E. Comparison of pathway enrichment in top loading genes of NMF_2 and NMF_3. Each data point is a pathway and the two axes are the −log(adjusted p-value) for the hypergeometric test for enrichment.

F. Comparison of pathway enrichment in top loading genes of NMF_4 and NMF_5. Each data point is a pathway and the two axes are the −log(adjusted p-value) for the hypergeometric test for enrichment.