SM2ST

We present SM2ST, a unified framework designed to integrate spatial transcriptomics and metabolomics. This framework leverages H&E images as a bridging modality for precise registration, employs a synergy of GANs and autoencoders for robust denoising and mapping, and incorporates the self-supervised super-resolution model, STMGraph, to establish a comprehensive spatial multi-omics analysis platform. adata_STMGrpah_pyG1_bruker_ITO_glass_15-30um_c_new.h5ad The 15µm SM data was downsampled to 30µm via average pooling for the mouse cerebral hemisphere. adata_STMGrpah_pyG1_bruker_ITO_glass_15-30-15um_c_new.h5ad The original 15µm data was downsampled to 30µm via average pooling and subsequently upsampled back to 15µm using STMgraph for the mouse hemispheric SM data. mouse_brain_30um1.csv mouse_brain_30umo1.csv ShinyCardinal-preprocessed 30µm MSI data of the mouse cerebral hemisphere. mouse_brain_15um1.csv ShinyCardinal-preprocessed 15µm MSI data of the mouse cerebral hemisphere. mice_brain_SMres20_align.h5ad Alignment of the 20µm mouse brain MSI data with the H&E-stained section was achieved using SM2ST. adata_SMLED_SMres20_c1_wmse1.h5ad Mapping of the 20µm mouse brain MSI data to the spatial transcriptomics (ST) data was achieved using SM2ST. adata_STMGrpah_pyG1_ito_30um_mad_rex_h.h5ad The original 30µm data was upsampled to 15µm using STMgraph for the mouse hemispheric SM data. Y7_T_adata_SMLED.h5ad Results of the Y7_T ccRCC data after SM2ST alignment and mapping.