Protocol for reference-free single-cell-resolution deconvolution of spot-based spatial transcriptomics with CoexpressDeconvolve

Spot-based spatial transcriptomics captures whole-transcriptome signal per spot but mixes multiple cells. Here, we present a protocol for using CoexpressDeconvolve, a reference-free approach that reconstructs single-cell-resolution, whole-transcriptome profiles directly from spot-based spatial transcriptomics data. We describe steps for housekeeping-gene cell-density calibration, joint overdispersed-gene selection, a gene co-expression manifold, and automated topic-number selection. We then detail procedures for latent Dirichlet allocation, hierarchical Bayesian sampling, and exporting to a 10×-compatible object for standard single-cell and spatial cell-cell communication analysis. For complete details on the use and execution of this protocol, please refer to Perik-Zavodskaia et al.¹.