FIG. 6.

Visualization of DestVI's computation workflow for spot deconvolution. DestVI uses information from both data modalities of the ST data (coordinates and scRNAseq). DestVI defines two latent variable models (LVMs) for each data modality: an LVM for modeling scRNAseq data (scLVM, shown at the top) and one that aims to model the ST data (stLVM, shown at the bottom). We describe each one in Sec. IV C. This figure was recreated for this manuscript based on illustrations from Lopez et al.⁶⁹